Computer Hardware Engineers
17-2061.00Median wage $161,740/yr76,660 employed (US)Rank #257 of 923 scored · top 28% by substitution
Research, design, develop, or test computer or computer-related equipment for commercial, industrial, military, or scientific use. May supervise the manufacturing and installation of computer or computer-related equipment and components.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution35
Exposure30
Augmentation66
Substitution — the headline: capability discounted by cost, barriers and adoption.
Exposure — technical capability alone, regardless of whether anyone deploys it.
Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.
Tasks on the substitution scale
18 rated tasks, binned by substitution score.
Position among all scored occupations
Distribution of 923 occupation scores; the marker is this occupation.
Tasks with substitution ≥ 70
6%
Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.
Why this score
The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.
Task automatabilityw 35%30
panel mean rating 2.2/5 → substitution pressure 30/100
Technical feasibility todayw 20%29
panel mean rating 2.2/5 → substitution pressure 29/100
Cost vs. human wagew 15%32
panel mean rating 2.3/5 → substitution pressure 32/100
Adoption barriersw 20%inverted — strong barriers lower the score50
panel mean rating 3.0/5 (barrier strength) → substitution pressure 50/100
Sector adoption velocityw 10%42
panel mean rating 2.7/5 → substitution pressure 42/100
Task breakdown (18 tasks)
Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.
Store, retrieve, and manipulate data for analysis of system capabilities and requirements.
71CI 55–86 · exposure 67 · augmentation 75 · importance 3.7/5 · click for rater detail
Store, retrieve, and manipulate data for analysis of system capabilities and requirements.
71| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Tech and engineering sectors have rapidly and deeply adopted automated data pipeline tools, cloud infrastructure, and AI-assisted data manipulation—this is among the fastest-adopted automation in software and hardware development. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Hardware engineering firms adopt AI tools for data analysis at a middling pace—faster than pure manufacturing but slower than software-only sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI tools significantly assist engineers in querying, visualizing, and organizing system data, raising their productivity in analysis; humans typically remain engaged for interpretation and decision-making. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly speeds up data querying, transformation, and preliminary analysis, letting engineers focus on interpretation and requirement tradeoffs. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Data storage, retrieval, and manipulation are core strengths of current AI systems and database tools. AI can automate most of these operations at scale with high speed and consistency, easily exceeding the 50% time-saving threshold, though some human judgment on data schema design and analysis goals may still be needed. |
| Task automatability | claude-sonnet-5 | 3/5 | Data storage, retrieval, and manipulation via scripts/queries is highly automatable, but tying this to meaningful analysis of system capabilities and requirements still requires engineering judgment and domain context. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers prevent automation; some organizational friction exists around data governance and quality assurance, but no licensed human signoff is required for the task itself. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement, but engineering sign-off and validation against hardware specs create moderate organizational friction and error-cost concerns. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated data pipelines and cloud-based storage/retrieval are orders of magnitude cheaper per operation than paying a hardware engineer's loaded wage to manually handle the same volume of data manipulation. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply handle bulk data manipulation, but integration with proprietary hardware specs and engineer oversight keeps overall cost comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature, production-grade systems (SQL databases, cloud storage platforms, Python/R data manipulation libraries, and AI-assisted ETL tools) reliably perform these tasks at enterprise scale across countless organizations daily. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Tools like SQL copilots, data pipelines, and AI-assisted analytics products exist and are used in production, but end-to-end reliable automation of requirements-linked analysis is narrower and less mature. |
Provide training and support to system designers and users.
64CI 41–87 · exposure 58 · augmentation 88 · importance 3.5/5 · click for rater detail
Provide training and support to system designers and users.
64| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Technical organizations and IT departments are rapidly deploying AI-driven documentation, chatbots, and self-service support; adoption is far advanced in information-sector companies where hardware engineers work. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Engineering and tech sectors are moderately fast adopters of AI tools for documentation and support, though specialized hardware training remains less digitized than software domains. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI excels at augmenting trainers and support engineers by instantly generating course materials, answering common questions, retrieving documentation, and freeing humans to focus on complex problem-solving and personalized instruction. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly augment training materials, generate documentation, answer common questions, and support troubleshooting, meaningfully boosting engineer productivity while humans remain central to complex instruction. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | AI systems today can generate instructional materials, tutorials, documentation, and provide real-time technical support via chatbots and Q&A systems, delivering training content at scale with significant time savings. Knowledge management and support ticket triage are effectively automatable with current LLMs and agent frameworks. |
| Task automatability | claude-sonnet-5 | 2/5 | Training and support involve interactive teaching, troubleshooting, and adapting to individual user needs, which AI can assist but not fully replace end-to-end at equal quality yet.assist |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No licensing requirement or legal mandate for an AI to provide training and support; the only friction is organizational preference for human touch in sensitive contexts and potential customer expectations, but these are soft barriers easily overcome. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement blocks AI from providing training/support, though organizational preference for expert human trainers and complexity of hardware systems create some friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered training systems and support bots cost a fraction of dedicated human instructors or support staff once deployed; the inference and infrastructure cost per interaction is orders of magnitude lower than loaded labor. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can reduce costs for tier-1 support and documentation, but complex hardware training still requires expert human time, making cost savings moderate rather than dramatic. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (technical documentation generators, chatbots, help desk automation) reliably handle much of training delivery and first-line support in production at scale, though high-complexity or deeply context-dependent support scenarios still benefit from human oversight. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Chatbots and documentation tools exist for basic support, but deployed products don't reliably deliver hands-on technical training for hardware engineering topics at scale. |
Test and verify hardware and support peripherals to ensure that they meet specifications and requirements, by recording and analyzing test data.
54CI 30–77 · exposure 50 · augmentation 63 · importance 3.9/5 · click for rater detail
Test and verify hardware and support peripherals to ensure that they meet specifications and requirements, by recording and analyzing test data.
54| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Hardware manufacturing and tech companies have been automating test suites and data analysis for decades; modern AI acceleration of this process is now standard in electronics and semiconductor sectors. Adoption of automated testing and ML-based anomaly detection is deep and rapid in high-digitization industries. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering and semiconductor/electronics sectors are moderate adopters of AI for test automation and predictive analytics, but adoption in physical verification labs lags behind software/professional services sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists by filtering noise from test data, flagging anomalies for engineer review, and predicting failure modes, raising the speed at which engineers can interpret results. However, the human role is already shifting toward oversight rather than data collection, so augmentation gains are incremental at this stage. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can meaningfully assist by rapidly analyzing large test datasets, flagging anomalies, and generating draft reports, letting engineers focus on interpretation and decision-making. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI and automated systems can execute standardized test protocols, collect vast amounts of sensor data, and perform statistical analysis of results against specifications with high efficiency. However, some aspects of interpreting unexpected failures or designing new tests may still benefit from human expert judgment, preventing a perfect 5. |
| Task automatability | claude-sonnet-5 | 2/5 | Some test data analysis and report generation can be automated, but designing test plans, interpreting anomalies, and validating hardware against specs requires physical setup, instrumentation, and engineering judgment that current AI cannot fully replace. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No licensing requirement mandates human oversight, and manufacturers are actively replacing manual test phases with automation. Customer preference for human verification is minimal in hardware testing; the main friction is system integration cost and legacy infrastructure inertia, not regulatory or legal bars. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement, but liability for hardware failures, safety certifications, and organizational quality processes create meaningful friction against full automation of sign-off. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Automated testing hardware and AI-driven analysis cost significantly less per test cycle than human labor, especially at scale. A single automated test rig can run hundreds of cycles while a human tester performs a fraction of that work, making the cost ratio heavily favorable for automation. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Physical test rigs, instrumentation, and oversight costs remain high; AI mainly reduces data analysis time but doesn't replace the capital and labor cost of physical hardware testing infrastructure. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed automated test systems, CI/CD pipelines, and data analysis tools are in production at major hardware firms and already handle much of this task reliably. AI-assisted anomaly detection and trend analysis are mature, though full end-to-end automation without human oversight remains imperfect in practice. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Deployed products exist for automated test equipment (ATE) and data logging, and AI-assisted anomaly detection is used in some labs, but end-to-end verification against specifications is still primarily human-driven with narrow AI point tools. |
Analyze user needs and recommend appropriate hardware.
49CI 38–61 · exposure 38 · augmentation 75 · importance 3.6/5 · click for rater detail
Analyze user needs and recommend appropriate hardware.
49| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Tech industry companies, cloud providers, and large retailers have already deployed AI-assisted or autonomous hardware recommendation systems at scale. Production adoption is visible in configuration tools, chatbots, and enterprise IT solutions across competitive tech sectors. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Hardware engineering and IT sectors are moderately adopting AI copilots for technical documentation and specs, but full-cycle needs analysis automation is still in pilot stages. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistants can significantly boost hardware engineer productivity by instantly cross-referencing compatibility databases, performance benchmarks, cost comparisons, and customer requirements, allowing engineers to focus on validation, trade-off analysis, and exceptional cases rather than routine lookup and comparison work. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can effectively assist by synthesizing user requirements, comparing specs, and drafting recommendations, significantly speeding up the engineer's research and analysis phase. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can perform significant parts of this task (gathering requirements, matching specs to common use cases, generating recommendations) but typically requires human expertise for complex or novel edge cases and final sign-off. Achieving 50% time savings with equal quality is realistic for routine recommendations but harder for specialized configurations. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires understanding nuanced business requirements, budget constraints, and technical tradeoffs specific to a user's context, which AI can support but not fully replace end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While recommendations may be followed by customer choice or junior staff oversight, there are no strict legal or regulatory requirements for a licensed professional to personally recommend hardware in most contexts. Organizational inertia and customer preference for human expertise provide modest friction but not hard legal barriers. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement, but organizational trust, procurement processes, and accountability for hardware failures create moderate friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven recommendation engines have minimal inference cost compared to the loaded salary of hardware engineers who would otherwise conduct detailed analysis. Once systems are in place, marginal cost per recommendation is very low relative to human labor equivalents. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate draft recommendations, but the human engineer's validation, stakeholder discussions, and integration testing keep overall costs closer to human-level for reliable outcomes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Products exist that can profile hardware needs and suggest configurations (configurators, chatbots with hardware databases), but they often struggle with nuanced requirements, industry-specific constraints, and uncommon architectures. Deployment is common in retail/e-commerce but reliability gaps remain in enterprise scenarios. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some AI-based configurator and recommendation tools exist for narrow hardware selection tasks, but no deployed product reliably conducts full needs analysis and recommendation across diverse enterprise hardware scenarios. |
Write detailed functional specifications that document the hardware development process and support hardware introduction.
40CI 30–50 · exposure 33 · augmentation 75 · importance 4.0/5 · click for rater detail
Write detailed functional specifications that document the hardware development process and support hardware introduction.
40| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Tech and hardware firms are piloting AI writing assistants and code-generation tools, but most still rely on humans for specification authorship. Adoption is growing in high-digitization sectors but remains cautious due to quality and compliance concerns. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering firms are generally slower AI adopters compared to software/IT sectors, though AI writing assistants are creeping into technical documentation workflows at a modest pace. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistants (LLMs, code templates, document generators) meaningfully boost engineer productivity by drafting boilerplate sections, suggesting structure, and helping organize complex requirements. Engineers stay in control while AI handles routine documentation work, significantly accelerating the overall writing process. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI is well-suited to accelerate drafting, formatting, consistency-checking, and summarizing technical specifications, significantly boosting engineer productivity even though human expertise remains essential for accuracy and completeness. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can draft sections of functional specifications and suggest structure, the task requires deep domain expertise, architectural decisions, and understanding of complex system integration that demands human judgment. Current AI tools lack the contextual knowledge to write complete, accurate hardware specifications end-to-end without substantial expert review and revision. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft substantial portions of functional specification documents from engineer inputs, notes, and prior specs, but requires deep domain understanding of the actual hardware design and iterative human validation to be accurate.detail.rich technical content.correctly.correct.the final document must accurately reflect real engineering decisions.this limits full automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While no explicit legal mandate requires a human signature, organizational quality standards, design review boards, and liability concerns create meaningful friction. Customers and internal processes typically require senior engineer sign-off on functional specifications, creating practical adoption barriers. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates a human author, but organizational IP sensitivity, proprietary design details, and internal review/sign-off processes create moderate friction against pure AI authorship. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted drafting may reduce initial writing time, but extensive expert review, technical validation, and rework are still required, keeping total cost close to or exceeding the cost of direct human specification writing by experienced engineers. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI drafting can reduce time spent on boilerplate and formatting, but the necessary human review, verification against actual hardware behavior, and iteration keep costs roughly comparable to a skilled engineer's efficient documentation process. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Products exist (code-generation tools, LLMs) that can generate specification outlines and fragments, but no deployed system reliably produces hardware functional specifications of production quality without expert oversight. Error rates remain high enough that these tools function as assistants rather than autonomous performers. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | LLM-based drafting tools and technical writing copilots exist and are used for documentation support, but there is no mature deployed product that reliably generates accurate hardware functional specs without heavy engineer review and correction. |
Monitor functioning of equipment and make necessary modifications to ensure system operates in conformance with specifications.
39CI 28–51 · exposure 38 · augmentation 75 · importance 3.4/5 · click for rater detail
Monitor functioning of equipment and make necessary modifications to ensure system operates in conformance with specifications.
39| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Large tech and cloud companies have deployed automated monitoring and some self-healing systems; however, adoption remains mixed across the broader hardware engineering sector, with many organizations still relying on manual oversight due to risk aversion and legacy infrastructure. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Hardware engineering sectors are adopting AI analytics and predictive maintenance tools at a moderate pace, though full autonomous monitoring-and-modification workflows are still uncommon. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-driven monitoring dashboards, predictive alerts, and diagnostic suggestions substantially augment human engineers' ability to spot issues early and prioritize fixes; engineers remain in control of modifications, but AI accelerates their decision-making and reduces time spent on routine detection. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-driven monitoring dashboards, anomaly detection, and simulation tools significantly enhance an engineer's ability to spot issues and evaluate potential fixes faster. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | Monitoring and diagnostics of equipment can be substantially automated through logs, sensors, and anomaly detection; however, determining root causes and deciding which modifications to make often requires domain expertise and judgment that current AI struggles with reliably in complex systems. |
| Task automatability | claude-sonnet-5 | 2/5 | Monitoring equipment functioning and making physical/design modifications requires hands-on diagnosis and hardware-level intervention that current AI cannot perform end-to-end, though data analysis portions can be assisted.rate accordingly.The core corrective action remains human-driven. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Hardware modifications carry high liability risk (system downtime, data loss, equipment damage), regulatory compliance often requires documented human authorization, and many enterprises mandate that qualified engineers sign off on changes—creating strong legal and organizational friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement typically applies, but engineering sign-off and safety/liability considerations for hardware changes create moderate organizational friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Monitoring infrastructure (sensors, dashboards, log aggregation) has been commoditized, but the human expert needed to interpret alerts and authorize modifications is still necessary and costly; AI handles part of the workload but does not yet replace the core decision-maker. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI-based monitoring tools add value but still require engineer oversight and physical modification labor, so the all-in cost is not substantially cheaper than human-led processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Commercial monitoring tools (dashboards, alert systems, basic anomaly detection) are mature and deployed; however, autonomous decision-making on modifications remains limited—most systems require human sign-off on significant changes, and edge cases still demand expert intervention. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Deployed monitoring/analytics tools exist for anomaly detection in hardware telemetry, but the modification/troubleshooting step is not handled reliably by any production AI system today. |
Provide technical support to designers, marketing and sales departments, suppliers, engineers and other team members throughout the product development and implementation process.
34CI 30–38 · exposure 25 · augmentation 63 · importance 3.7/5 · click for rater detail
Provide technical support to designers, marketing and sales departments, suppliers, engineers and other team members throughout the product development and implementation process.
34| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption of AI for technical support across product development teams remains limited despite digitization; companies tend toward augmentative tools (chatbots for FAQs) rather than autonomous technical support agents, reflecting skepticism about AI judgment in high-stakes cross-functional roles. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Engineering and tech sectors are adopting AI copilots and internal knowledge tools at a moderate pace, with pilots for documentation and troubleshooting support becoming common but not yet deeply embedded. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by drafting technical documentation, summarizing stakeholder feedback, flagging compatibility issues, and preparing background materials for technical support staff. However, the human engineer must remain in the loop for final decision-making and stakeholder communication on complex matters. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can meaningfully speed up drafting technical explanations, summarizing specs, answering routine queries, and preparing materials for marketing/sales, letting engineers focus on complex judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | This task requires real-time collaborative communication, contextual understanding of diverse stakeholder needs, and dynamic problem-solving across multiple departments. While AI can draft some technical documentation or provide template responses, it cannot reliably handle the full scope of cross-functional support, stakeholder management, and nuanced technical guidance that current AI systems perform well. |
| Task automatability | claude-sonnet-5 | 2/5 | This task spans varied ad hoc communication, troubleshooting, and cross-functional problem-solving that requires deep contextual knowledge of specific hardware designs, making full automation unlikely though some Q&A elements could be assisted. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | There are moderate adoption barriers: organizational trust in AI for cross-functional communication, liability concerns if incorrect technical guidance affects product timelines or quality, and stakeholder preference for human judgment on sensitive design and implementation decisions. However, no strict regulatory barrier legally prevents AI deployment. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement, but organizational trust, accountability for technical decisions, and the relationship-based nature of cross-team support create moderate friction against full substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI assistance for technical support requires significant human oversight to ensure accuracy and appropriateness for diverse stakeholders, limiting cost savings. Integration with existing development workflows and quality assurance add overhead that narrows the cost advantage over experienced technical support staff. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI assistance is cheap per query, the need for a human engineer to remain in the loop for judgment calls and specialized knowledge means overall cost savings are modest, not order-of-magnitude. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs end-to-end technical support across design, marketing, sales, and engineering teams with the interpersonal and contextual sophistication required. Chatbots and AI support tools exist but have high error rates in complex multi-stakeholder scenarios and struggle with the judgment calls needed in product development. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI chatbots and copilots can answer some technical questions or draft documentation, but no deployed product reliably substitutes for an engineer's live, context-aware technical support across teams and suppliers. |
Specify power supply requirements and configuration, drawing on system performance expectations and design specifications.
30CI 30–30 · exposure 25 · augmentation 63 · importance 3.3/5 · click for rater detail
Specify power supply requirements and configuration, drawing on system performance expectations and design specifications.
30| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Hardware engineering remains relatively conservative and manual; while some firms pilot AI-assisted design tools, widespread production adoption is still limited due to the high cost of errors and the distributed, specialized nature of hardware engineering teams. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering firms are slower AI adopters compared to software/IT sectors, with AI tools mostly used for auxiliary tasks like documentation or simulation rather than core specification decisions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by rapidly retrieving power consumption benchmarks, generating candidate configurations, and cross-checking arithmetic, helping engineers iterate faster; however, the task's dependence on system-specific knowledge and risk assessment limits transformative impact. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by running simulations, suggesting component specifications, checking against datasheets, and flagging design inconsistencies, boosting engineer throughput on this task. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can extract and process power specifications from design documents and suggest configurations based on standard parameters, the task requires deep judgment about system performance expectations, thermal profiles, and edge cases that current AI systems struggle with reliably. Only fragments of the task (parameter lookup, basic calculations) are easily automated. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires integrating system-level performance requirements, thermal/electrical constraints, and design tradeoffs that current AI can assist with but not reliably execute end-to-end without significant engineer oversight.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Specifications must ultimately be signed off by a licensed engineer or architect; liability for power failures or system instability creates organizational friction and requires human professional accountability, though the task itself is not legally restricted to licensed professionals in all jurisdictions. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing requirement, but engineering sign-off, safety/reliability certification processes, and organizational review workflows create meaningful friction against pure AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI tools (via LLMs or assistants) have low inference cost, but integration overhead, validation, and the need for human expert review to catch errors makes the all-in cost comparable to or higher than a competent engineer's hourly rate for this task. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Engineering judgment, validation against real system constraints, and liability for errors mean human engineer time still dominates; AI tools reduce but don't eliminate the cost of this specialized task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs full power supply specification end-to-end; tools exist for narrowly scoped calculations or lookup (power estimators), but they require expert human validation and cannot independently navigate the interdependencies between system performance, thermal management, and actual hardware constraints. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some EDA and design tools offer power estimation and recommendation features, but no deployed product autonomously specifies full power supply requirements for novel hardware designs at production quality. |
Analyze information to determine, recommend, and plan layout, including type of computers and peripheral equipment modifications.
30CI 30–30 · exposure 25 · augmentation 63 · importance 2.8/5 · click for rater detail
Analyze information to determine, recommend, and plan layout, including type of computers and peripheral equipment modifications.
30| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Hardware engineering and infrastructure planning remain concentrated in specialized IT and engineering teams in larger organizations, with slower digitization and automation adoption compared to information-heavy professional services. Real production deployment of autonomous planning systems is uncommon. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering sits in a moderately digitized but physically-grounded sector; AI adoption for actual layout decisions remains in pilot/research stages rather than production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist by analyzing equipment specifications, generating candidate layouts, and summarizing cost–performance trade-offs, raising engineer productivity in research and early-stage planning. However, the engineer remains responsible for validating recommendations against organizational context and constraints. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can meaningfully assist by synthesizing specs, generating design options, and flagging compatibility issues, substantially speeding the analysis phase even though a human finalizes the plan. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist in analyzing hardware specifications and generating layout recommendations based on given constraints, the task requires significant domain expertise, understanding of organizational context, and integration with existing systems that current AI handles poorly. End-to-end execution would need substantial human oversight and revision, falling well short of the 50% time-saving bar. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires synthesizing technical specs, cost, performance and integration constraints into a recommended hardware layout, which involves judgment and physical/engineering constraints AI cannot fully verify today.5-star automation isn't reached; AI can assist analysis but not fully replace the planning decision. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Some organizational friction and need for human sign-off exist, but no strong legal or licensing barriers prevent automation or assist-only deployment. Industry practice and liability concerns create moderate friction rather than hard prohibitions. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement blocks AI use, but organizational risk aversion, safety-critical hardware constraints, and need for engineering sign-off create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The overhead of integration, validation, and human expert review to ensure correctness likely makes AI assistance cost-comparable to or exceeding the loaded wage of a skilled hardware engineer for this judgment-heavy task. Significant human oversight is required. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Given the need for human engineering validation, simulation, and vendor-specific knowledge, AI reduces some analysis time but doesn't yet replace the labor cost of a qualified engineer at scale. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No mature deployed products reliably perform the full scope of analyzing organizational needs, recommending hardware architectures, and planning equipment layouts independently. LLMs can draft recommendations but lack the reliability for production deployment without expert human review and domain validation. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | There are no widely deployed production systems that autonomously perform full hardware layout planning and equipment recommendation; existing tools are decision-support only, not autonomous planners. |
Recommend purchase of equipment to control dust, temperature, and humidity in area of system installation.
30CI 25–35 · exposure 20 · augmentation 63 · importance 2.7/5 · click for rater detail
Recommend purchase of equipment to control dust, temperature, and humidity in area of system installation.
30| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Hardware engineering and data center operations have moderate digitization but slow AI adoption for equipment procurement decisions, as these remain human-driven in most organizations due to risk aversion and the need for accountability. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering and facilities planning sectors have slower AI adoption for physical infrastructure decisions compared to pure information-based tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist by quickly compiling equipment specifications, comparing environmental standards, and surfacing options—helping engineers make faster, better-informed recommendations while the engineer retains responsibility for final selection. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can effectively assist by researching equipment specifications, comparing options, calculating environmental requirements, and drafting recommendation reports, significantly speeding up the human's research and documentation process. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can analyze environmental specifications and suggest equipment based on technical parameters, this task requires nuanced judgment about site-specific conditions, cost-benefit tradeoffs, and integration with existing infrastructure—factors that typically demand human expertise and on-site assessment. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires site-specific engineering judgment about environmental control needs based on equipment specs and physical installation conditions, which AI cannot directly observe or assess without significant human-provided data.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | This task typically falls within the responsibility of licensed or certified engineering roles in regulated environments; organizations often require human engineers to sign off on equipment recommendations for liability and compliance reasons, creating moderate adoption friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No strict licensing requirement for this specific recommendation task, though engineering sign-off and liability for equipment failures create some organizational caution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference for equipment recommendation is cheap, but the task's value lies in expert judgment and customization; the loaded cost of a qualified hardware engineer doing this work exceeds what AI alone can deliver due to required validation and site-specific analysis. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI could assist with research and drafting recommendations but a human engineer must still evaluate site conditions and finalize purchase decisions, limiting cost savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI systems can retrieve and summarize equipment options from specifications databases, but no deployed product reliably recommends control equipment for complex, context-dependent installation environments without significant human verification and site surveys. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously assesses physical installation sites and recommends specific environmental control equipment purchases; this remains a human engineering judgment task. |
Update knowledge and skills to keep up with rapid advancements in computer technology.
29CI 19–40 · exposure 17 · augmentation 75 · importance 4.3/5 · click for rater detail
Update knowledge and skills to keep up with rapid advancements in computer technology.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Tech companies are beginning to use AI for content discovery and skill recommendations, but adoption remains uneven; many engineers still rely on self-directed learning through conferences, courses, and peer networks rather than AI-driven continuous learning systems. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Engineers in tech-heavy fields rapidly adopt AI tools (chatbots, summarizers, personalized learning platforms) to keep pace with new technology, consistent with high digitization adoption patterns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments this task by filtering and summarizing the vast volume of new research and technologies, personalizing learning paths, and automating detection of emerging trends—allowing engineers to focus their study time more efficiently while remaining in control of actual skill development. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly augments this task by rapidly summarizing new research, generating explanations, and answering technical questions, greatly speeding up how engineers learn and stay current. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires self-directed learning, judgment about which technologies are relevant, and integration of knowledge into existing skill sets—activities that depend on human motivation, career planning, and contextual understanding that current AI cannot independently execute end-to-end. |
| Task automatability | claude-sonnet-5 | 2/5 | Continuous learning is an ongoing personal/professional development activity that AI can support with information curation but cannot substitute for the human process of internalizing and applying new knowledge.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Professional credibility, licensure in some contexts, and peer recognition depend on demonstrated mastery that only humans can claim; organizational cultures also value direct evidence of engineer engagement with emerging tech, not proxy metrics from AI-assisted discovery. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal barrier prevents AI assistance in learning, though professional certification and hands-on skill validation still require human demonstration of competence. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools for content curation and summarization are inexpensive, but the task fundamentally requires paid human time to engage with material, practice, and validate competency—making the total cost per outcome still heavily weighted toward human labor. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI-assisted research and summarization is cheap, but the actual skill acquisition still requires human time and practice, so the cost comparison is not favorable to full automation. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can curate technical content, summarize research papers, and recommend learning resources, no deployed product actually performs the full task of a professional staying current through independent self-management and skill validation. AI assists but does not replace the human learning process. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools (search, summarization, newsletters) exist to help surface new information, but no product autonomously 'updates the knowledge and skills' of an engineer in a deployable, complete sense. |
Design and develop computer hardware and support peripherals, including central processing units (CPUs), support logic, microprocessors, custom integrated circuits, and printers and disk drives.
29CI 25–32 · exposure 25 · augmentation 63 · importance 4.2/5 · click for rater detail
Design and develop computer hardware and support peripherals, including central processing units (CPUs), support logic, microprocessors, custom integrated circuits, and printers and disk drives.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Hardware engineering is concentrated in specialized firms with long product cycles, established design teams, and high switching costs. Adoption of AI-assisted design tools is incremental and measured; full displacement of hardware engineers through automation is not occurring in production. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Semiconductor and hardware firms are piloting AI in chip design (e.g., Google's use of RL for chip floorplanning) but widespread production deployment across the full hardware design lifecycle remains uneven and specialized. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist with specific subtasks—circuit simulation, layout suggestions, component research, design rule checking—improving engineer productivity on portions of the design workflow. However, the assistance is narrowly scoped to well-defined steps within a larger creative and integrative process. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids engineers via simulation, design-space exploration, verification, and code/HDL generation, meaningfully speeding iteration cycles while humans retain final design authority. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with circuit design, simulation, and component selection, the task requires creative system-level design decisions, trade-off analysis between performance/power/cost, and iterative prototyping that demand human expertise. Current AI cannot independently produce production-ready hardware designs meeting complex specifications end-to-end. |
| Task automatability | claude-sonnet-5 | 2/5 | AI (EDA tools, generative design aids) can assist portions of hardware design like layout optimization or code generation for HDL, but core architectural design, tradeoff analysis, and physical validation still require deep human engineering judgment and cannot be fully automated end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Hardware design involves patent creation, safety certifications (FCC, UL), liability for product defects, and regulatory compliance. Organizations and clients typically require human engineers to own and sign off on designs; liability asymmetry and institutional risk aversion create strong adoption friction. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement for hardware engineers akin to medicine/law, but high liability for design failures, IP protection concerns, and deep organizational reliance on validated engineering processes create real friction against full AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Hardware design requires expensive specialized tools (CAD, simulation software, prototyping), domain expertise, and iterative cycles. AI-assisted tools reduce some design time but cannot eliminate the need for skilled engineers; the all-in cost of AI workflows remains comparable to or higher than human-led design. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Specialized AI design tools have high licensing, compute, and integration costs, and still require expert engineers to interpret and validate outputs, making all-in costs comparable to or higher than dedicated engineering teams for novel designs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI tools exist for circuit simulation, schematic generation, and design optimization, but no deployed product autonomously designs and develops complete hardware systems at scale. Existing tools require significant human oversight, domain knowledge, and manual intervention to produce viable results. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some AI-assisted EDA tools (e.g., for chip placement/routing) are deployed in production at firms like Google/Synopsys, but full hardware design workflows for CPUs and peripherals remain human-led with AI as a narrow tool, not a reliable end-to-end performer. |
Confer with engineering staff and consult specifications to evaluate interface between hardware and software and operational and performance requirements of overall system.
29CI 25–32 · exposure 25 · augmentation 63 · importance 4.1/5 · click for rater detail
Confer with engineering staff and consult specifications to evaluate interface between hardware and software and operational and performance requirements of overall system.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Hardware engineering remains a domain with slower AI adoption compared to software-only roles; while firms are exploring AI-assisted design review, production-level replacement of interface evaluation and staff consultation is rare. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Engineering and tech sectors are moderately fast adopters of AI copilots for documentation and code review, but the specific meeting/consultation activity remains largely unautomated in practice. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by analyzing specifications, flagging potential compatibility issues, and summarizing technical documentation, helping engineers review options faster, though the engineer remains in the decision-making loop. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by parsing specifications, flagging interface mismatches, and drafting technical summaries ahead of or during human discussions, boosting engineer efficiency. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | This task requires interpreting specifications, making judgment calls about interface compatibility, and consulting with human engineering staff—activities that involve negotiation, nuance, and domain-specific reasoning. Current AI can assist in analyzing specs or summarizing requirements but cannot independently evaluate system interfaces and performance requirements at the level required for end-to-end task completion with 50% time savings. |
| Task automatability | claude-sonnet-5 | 2/5 | This task centers on live cross-functional discussion, negotiation, and integration of tacit engineering judgment across teams, which current AI cannot conduct autonomously end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Professional responsibility for system-level interface decisions typically rests with licensed or senior engineers who must sign off on hardware-software integration. Organizational practice, liability concerns, and the need for human accountability create significant adoption barriers. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement, but organizational reliance on engineer expertise, accountability for system failures, and need for real-time technical judgment create moderate friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The oversight, human review, and integration burden for AI-assisted evaluation would likely match or exceed the cost of direct human consultation, since the task requires judgment calls that cannot be outsourced to cheap inference alone. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Human engineers with domain context remain necessary for meaningful consultation; AI assistance reduces some prep time but doesn't replace the core wage cost of the collaborative evaluation. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can process technical specifications and suggest compatibility issues, no deployed product reliably performs the full task of conferring with staff and making authoritative interface evaluations in production engineering environments. This requires real-time collaboration, architectural judgment, and accountability that deployed systems do not yet handle independently. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools can summarize specs and flag inconsistencies, but no deployed product reliably conducts the actual hardware-software interface evaluation and interdisciplinary conferring at production quality. |
Build, test, and modify product prototypes, using working models or theoretical models constructed with computer simulation.
29CI 25–32 · exposure 25 · augmentation 75 · importance 4.1/5 · click for rater detail
Build, test, and modify product prototypes, using working models or theoretical models constructed with computer simulation.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Simulation and CAD-assisted design are common in hardware firms, but actual automation of the full prototype build-test-modify cycle remains limited; adoption is selective and domain-dependent, with many firms still relying on traditional iterative human-led prototyping. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering sectors adopt AI tools for simulation and design assistance at a moderate pace, but physical prototyping workflows remain slower to digitize compared to pure information-work sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered simulation, modeling, and design optimization substantially accelerate the engineer's ability to explore design alternatives and predict behavior, making the human engineer far more productive even when they remain directly involved in physical testing and iteration. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-driven simulation tools, generative design, and rapid modeling significantly speed up the theoretical/simulation portion of prototype development, meaningfully boosting engineer productivity even though physical build/test stays human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with simulation and modeling, building and physically testing hardware prototypes requires hands-on construction, physical assembly, and real-world validation that cannot be fully automated end-to-end by current AI systems without substantial human intervention and oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with simulation setup, design suggestions, and analysis but physically building and testing hardware prototypes requires human manipulation of physical materials and lab equipment that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Prototype development for safety-critical hardware faces substantial liability and regulatory oversight barriers; legal and professional responsibility for design validation typically cannot be fully delegated to automation, and client/stakeholder requirements often demand engineer sign-off. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement for this task itself, but physical safety testing, equipment access, and organizational validation processes create meaningful friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The cost of AI-driven simulation tools plus the still-necessary human engineering labor for physical prototype construction and validation remains comparable to or higher than direct human engineering work, particularly given the need for specialized hardware expertise. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Simulation software licenses plus compute can reduce some design iteration costs, but physical prototype construction and testing still requires skilled engineers and lab technicians, keeping overall costs comparable to human-driven processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI tools exist for simulation and computational modeling of prototypes, but no deployed products reliably handle the full cycle of physical building, testing, and iterative modification without significant human expertise and manual work guiding the process. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | EDA/CAD tools with AI-assisted features exist for simulation and design optimization, but no deployed product autonomously builds and physically tests prototypes; these remain human-driven workflows with AI as a co-pilot. |
Evaluate factors such as reporting formats required, cost constraints, and need for security restrictions to determine hardware configuration.
29CI 25–32 · exposure 25 · augmentation 63 · importance 3.6/5 · click for rater detail
Evaluate factors such as reporting formats required, cost constraints, and need for security restrictions to determine hardware configuration.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | While IT organizations are early-stage in adopting AI for infrastructure planning, the security-critical nature and accountability requirements mean current adoption remains experimental and oversight-heavy rather than autonomous deployment. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Engineering and IT sectors are moderately adopting AI copilots for technical decision support, but full delegation of configuration decisions remains rare. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by automatically extracting reporting format requirements, matching cost constraints to candidate hardware, and flagging security trade-offs, allowing engineers to focus on reconciling complex conflicts and validating business logic rather than routine documentation review. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can effectively assist by summarizing requirements, comparing cost/security tradeoffs, and generating configuration options, significantly speeding up the engineer's analysis. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can extract and analyze requirements from documents and suggest hardware configurations, the task involves interpreting ambiguous business constraints, applying nuanced judgment about security trade-offs, and cross-referencing multiple competing factors—activities that require human domain expertise and accountability that current AI cannot fully replicate end-to-end. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires synthesizing organizational requirements, cost tradeoffs, and security policy into a specific hardware configuration decision, which involves contextual judgment and accountability that current AI cannot fully replace end-to-end.rows.9K:1)Actually this requires nuanced tradeoff reasoning that AI can support but not autonomously decide with equal quality at scale.rows.9K:1) |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Security, compliance, and risk liability create strong organizational and sometimes regulatory barriers; hardware misconfiguration can expose systems to breach, and the engineer typically bears accountability for security-relevant choices that AI cannot legally or safely assume full responsibility for. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement, but security restrictions and cost accountability create organizational friction and liability concerns that discourage full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference for requirement extraction and configuration suggestion is cheap, but the task requires human engineers to review, validate, and take responsibility for the output, making total cost-per-configured-system competitive with or higher than direct human engineering. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Using AI to assist would require significant human oversight and validation given cost and security stakes, so all-in cost savings versus an engineer's judgment are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No production systems reliably perform independent hardware configuration determination based on real organizational constraints; AI tools exist to assist with documentation parsing and spec lookup, but humans must validate security implications, reconcile conflicting requirements, and sign off on final decisions. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools can help draft configuration options or summarize constraints, but no deployed product autonomously performs this full evaluative decision-making task in production. |
Select hardware and material, assuring compliance with specifications and product requirements.
28CI 25–30 · exposure 25 · augmentation 63 · importance 3.7/5 · click for rater detail
Select hardware and material, assuring compliance with specifications and product requirements.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Hardware engineering remains a specialty function in embedded and design-heavy sectors that have lagged broad AI adoption. Pilots of parametric selection tools exist, but production adoption remains limited; most organizations still rely on engineer expertise and vendor relationships. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering firms are traditionally slower AI adopters than software/professional services; AI tool use here is mostly limited to search/co-pilot assistance rather than production-scale agentic adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | LLMs and retrieval systems can assist by summarizing datasheets, cross-checking specifications against requirements, and flagging potentially violated constraints—useful augmentation that can improve engineer productivity. However, the engineer remains responsible for final validation and trade-off decisions. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can significantly speed up datasheet research, spec comparison, and compatibility checking, meaningfully augmenting engineers even though final selection and compliance sign-off remain human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Hardware selection involves evaluating complex interdependencies between performance specifications, power budgets, thermal requirements, cost constraints, and availability—requiring multi-dimensional trade-off reasoning. While AI can retrieve and filter component datasheets, current systems cannot reliably validate end-to-end compliance across the full constraint set or handle novel constraint combinations typical in hardware design. |
| Task automatability | claude-sonnet-5 | 2/5 | Component selection requires synthesizing datasheets, supplier constraints, cost, and system-level tradeoffs that current AI can partially support but not reliably decide end-to-end without engineer review., and no off-the-shelf system does this autonomously with equal quality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Hardware selection in regulated industries (automotive, aerospace, medical, defense) often requires explicit engineer sign-off and traceability documentation by license; liability for incorrect selection falls on the responsible engineer. Organizational and regulatory requirements make direct automation difficult. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No explicit licensing requirement, but liability for component failures, supply chain compliance (e.g., safety, RoHS, export control) and organizational sign-off processes create meaningful friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI systems capable of hardware selection would require significant integration with design tools, constraint databases, and supplier APIs, plus ongoing human oversight to catch missed requirements. Current overhead makes the all-in cost comparable to or exceeding the human engineer's time for routine selections. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply search datasheets and cross-reference specs, but the human engineering judgment, supplier negotiation, and compliance verification still dominate cost, keeping overall savings modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product systematically performs this task reliably end-to-end in production. CAD and parametric selection tools exist, but they require extensive manual setup and human validation of constraints; they do not autonomously assure compliance across all specifications. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | There are AI-assisted BOM tools and search/recommendation systems for parts, but no mature deployed product autonomously selects hardware/materials assuring full spec compliance in production engineering workflows. |
Assemble and modify existing pieces of equipment to meet special needs.
19CI 16–21 · exposure 16 · augmentation 50 · importance 2.9/5 · click for rater detail
Assemble and modify existing pieces of equipment to meet special needs.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption of autonomous assembly in custom hardware engineering is slow outside high-volume manufacturing. Most hardware engineers work in smaller teams or specialized contexts where bespoke modification is common, sectors with limited automation investment and preference for human technical expertise. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Hardware engineering involves physical prototyping and lab work, a sector with slower AI/robotics adoption compared to purely digital information work. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI tools can assist with design optimization, parts selection, and documentation of assembly procedures, raising engineer productivity in planning phases. However, AI's role remains peripheral to the hands-on assembly and troubleshooting work that defines the task. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with design simulations, generating modification plans, or troubleshooting guidance, but the physical assembly itself still requires human hands. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Assembling and modifying hardware requires physical dexterity, spatial reasoning, and real-time problem-solving in unstructured environments. Current AI systems lack embodied manipulation capabilities to reliably handle diverse equipment; while perception and planning components exist in research, end-to-end physical assembly with quality parity is not achievable at 50% time savings today. |
| Task automatability | claude-sonnet-5 | 2/5 | Physical assembly and modification of hardware equipment requires manual dexterity, tool use, and real-world manipulation that current AI systems cannot perform autonomously; only design/planning subcomponents are assistable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Physical modification of equipment often requires certification, warranty liability, and sign-off by licensed technicians, particularly in regulated industries (telecommunications, defense, medical). Customer acceptance of non-human assembly and the need for human inspection create substantial organizational and legal barriers. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement specifically blocks automation, but the need for physical dexterity, safety considerations, and hands-on customization creates strong practical friction against substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Custom robotic systems capable of assembly and modification are capital-intensive and require significant integration engineering. The loaded cost of such systems, including setup, programming, and maintenance, far exceeds the wage of a skilled hardware engineer performing this task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI-only substitute for physical assembly work, so any AI cost comparison is moot; human labor remains the only functional option for this physical task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed commercial products reliably perform physical hardware assembly and modification autonomously. Robotic solutions exist for narrow, repetitive tasks in controlled factory settings, but general-purpose assembly to meet custom needs remains research-stage and requires human oversight. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical hardware assembly/modification end-to-end; robotics for custom, non-standardized equipment modification remains research-stage or limited to controlled manufacturing lines. |
Direct technicians, engineering designers or other technical support personnel as needed.
4CI 0–7 · exposure 0 · augmentation 38 · importance 3.7/5 · click for rater detail
Direct technicians, engineering designers or other technical support personnel as needed.
4| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Even in high-tech sectors, team direction and personnel management remain firmly human functions; no measurable displacement of engineering managers or supervisors by AI agents has occurred in production environments. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While hardware engineering firms adopt AI tools for design tasks, adoption of AI for supervisory/managerial direction of personnel is essentially nonexistent. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can assist with scheduling, workload analysis, and documentation, but cannot augment the core leadership function meaningfully since judgment, authority, and accountability remain essential to the task. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help draft instructions, track task assignments, or summarize technical specs to support the engineer's direction of others, but it doesn't replace the interpersonal leadership itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires real-time human supervision, delegation decisions, and interpersonal judgment to manage technical personnel effectively. Current AI cannot autonomously direct teams, make personnel decisions, or provide leadership oversight in production environments. |
| Task automatability | claude-sonnet-5 | 1/5 | Directing people involves real-time leadership, judgment calls, and interpersonal accountability that current AI cannot perform end-to-end; AI cannot legitimately manage or direct human staff.9 |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task has hard organizational and legal barriers: management authority typically requires human accountability, organizational hierarchies mandate human supervisors, and liability for personnel decisions falls on licensed professionals. Customers expect human leadership. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Organizational structure, accountability, and liability for engineering decisions require a human supervisor; direction of staff is inherently tied to human authority and responsibility. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Implementing an AI system to manage technical teams would be vastly more expensive than retaining human supervisors, given integration costs, customization, and the requirement for human sign-off on all significant decisions. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this managerial function, so no meaningful cost comparison favors AI over the human engineer. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system reliably performs autonomous team leadership or personnel direction in production. While AI can draft instructions or assist with scheduling, actual direction of technical staff requires human authority and accountability that AI systems cannot provide. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously directs technical support personnel in engineering settings; this remains outside current AI product scope. |
Related occupations — Architecture & Engineering
How to read this
A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.
What would change this score
New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.