Humans in AI Territory: A Thought on AI Adoption Patterns

Author(s): Shashwat (Shawn) Gupta

Originally published on Towards AI.

TLDR

(from my X.com/LinkedIn post)

Recently read MIT’s “State of AI in Business 2025” Report, which reveals a telling disconnect: despite $35–40B invested, 95% of AI pilots fail to deliver business impact, yet 90% of workers use personal AI tools in a thriving “shadow AI economy”. Clearly, generative AI is in the ‘trough of disillusionment’ (Gartner 2025). An interesting question emerges: can we look for human-machine collaboration patterns from other industries where machines outperform humans?

While researching historical precedents, I found 4 paradigms particularly relevant:

1. 🔄 Dot-com Era : Initial chaos, then transformation with net job creation in new categories (**Jevon’s Paradox**) with humans adapting to emerging roles.

2. ✈️ Aviation: Autopilots handle precision and can function autonomously, pilots do strategic oversight with emergency-case handling (**whole system-supervision**)

3. 🛡️ Nuclear/Finance: Advanced automation with mandatory human oversight for decisions related to criticality and compliance. (**regulatory and safety-critical roles**)

4. 🚀 Medical Radiology: AI improves diagnosis 16%+, but doctors remain essential for clinical judgment (**accountability, agency and person-to-person context**)

This report dives deeper into each of the 4 paradigms touching upon 2 key reports for this
1. MIT 2025 Report (State of AI in Business 2025)
2. Gartner AI 2025 Hype Cycle

Introduction

Unlike previous technological shifts, AI is fundamentally different — it’s solving problems rather than just automating processes, which means it’s here to stay. The current wave of AI transformation is accelerating at unprecedented speed. ChatGPT was released about two years ago; OpenAI reports that usage now exceeds 300 million weekly users and that over 90 percent of Fortune 500 companies employ its technology [15]. The internet did not reach this level of usage until the early 2000s, nearly a decade after its inception.

When great minds in tech can’t predict the future with certainty, I certainly can’t. But the data is becoming clear: 14% of all workers have already been displaced by AI [5], and while you’re reading this article, 513 people lost their jobs to AI today. Yet this isn’t just a story of displacement — it’s about transformation. As an interesting thought experiment, examining parallel industries and historical trends can illuminate what work might look like in the coming decades.

In a post I made 1 year back, I compiled various reports on jobs in AI world(upto mid 2024) and discussed about the AI Hype (nothing new any new tech goes through the same phases). This is an addition to that where we draw parallels from history in industries which show similar characteristics to AI and assume that similar impact will be there. The link is : https://medium.com/towards-artificial-intelligence/towards-artificial-general-intelligence-agi-and-what-is-in-store-for-us-a-hype-story-3850bfe87ff8

I also talked about Technologies in AI that are at infancy here: https://medium.com/towards-artificial-intelligence/the-critical-nuances-of-todays-ai-and-the-frontiers-that-will-define-its-future-4ca9e073cd2d

Four Paradigms for Understanding Human Roles in the AI Era

1. 🔄 The Dot-Com Era Paradigm: Jevon’s Paradox in Action

Initial chaos, then transformation with net job creation in new categories

The dot-com boom of the late 1990s provides our framework for understanding AI’s trajectory through technological adoption cycles. During this period, unprecedented automation of business processes and global connectivity fundamentally changed how work gets done, but companies that survived didn’t just automate existing processes — they reimagined entire business models.

Industry Growth Numbers: The internet took nearly a decade after inception to reach widespread adoption in the early 2000s.

AI Parallel Growth: ChatGPT was released about two years ago and now exceeds 300 million weekly users with over 90% of Fortune 500 companies employing OpenAI technology [15]. 97 million jobs created globally due to AI (WEF forecast, 2025) while 85 million jobs displaced, resulting in a net gain of +12 million jobs [21]. AI and data science specialists are among the fastest-growing job categories in 2025 [12].

This pattern is repeating with AI. Like the dot-com era, we’re seeing initial speculation and rapid adoption, followed by more measured integration. The key insight here is Jevons’ Paradox: as AI makes certain tasks more efficient, demand for human workers with complementary AI skills actually increases. The 25% wage premium for AI-skilled workers rising even higher in 2025 demonstrates this counterintuitive effect [4].

The market value lost in the dotcom bust has been largely regained through the rise and maturation of successful tech companies, with today’s top tech firms representing around 40% of the S&P 500’s market capitalization, compared to about 47% of the S&P 500 being tech-related during the dotcom peak in 2000 [51,52]. While the dotcom bubble saw a speculative surge with many companies failing, survivors like Amazon and Google have grown into dominant giants with substantial real earnings, driving significant market value recovery [51]. Additionally, the number of software companies has greatly expanded far beyond the internet-focused startups of the dotcom era, reflecting broader technological advances and adoption across industries [53,54]. This expansion is evidenced by the proliferation of AI startups, cloud computing firms, and digital service providers, underscoring a transformative shift in the economy where software plays a central role in nearly all sectors today [55].

The lesson: massive technological shifts create initial turbulence but ultimately lead to sustainable transformation of entire economic sectors, often with net job creation in new categories.

2. ✈️ The Aviation Paradigm: Whole System Supervision

Autopilots handle precision, pilots do strategic oversight with emergency-case handling

Commercial aviation represents perhaps the most mature example of human-machine collaboration. Modern aircraft can handle virtually every aspect of flight with precision far exceeding human capabilities, yet pilot employment is booming, not declining.

Industry Growth Numbers: 546% increase in starting salaries for regional airline pilots since 2000, soaring from $16,000 to $108,000 annually by 2024. The median annual wage for airline pilots, copilots, and flight engineers was $226,600 in May 2024 [2,3]. Overall employment of airline and commercial pilots is projected to grow 5 percent from 2023 to 2033, with about 18,500 openings projected each year [2]. 660,000 new pilots needed worldwide by 2044, with 119,000 of these in North America [6].

AI Parallel Growth: Wage premium for AI skills comparing workers in the same job with and without AI skills has risen from 25% last year to higher premiums in 2025 [13]. 91% of companies using or planning to use AI in 2024 will hire new employees in 2025 [17].

The key insight: pilots have evolved from direct flight control to sophisticated system management and supervisory oversight. They monitor autopilot performance, respond to system alerts, make strategic decisions about route changes, and provide critical judgment during unexpected situations. Research from MIT’s Aeronautics Department in 2024 highlights AI’s limitations: while it excels within predictable parameters, it falters in unpredictable scenarios like engine failures during storms or emergency landings in restricted zones.

3. 🛡️ The Nuclear/Finance Paradigm: Regulatory and Safety-Critical Roles

Advanced automation with mandatory human oversight for critical decisions

This paradigm examines industries where regulatory, safety, or human-centric factors create enduring requirements for human involvement despite advanced automation capabilities.

Industry Growth Numbers — Nuclear: Nuclear electric power generation businesses employed 58,517 workers in 2023, 2.8% more than in 2022, with the largest number of new jobs (1,079) in professional and business services [41]. However, overall employment of power plant operators is projected to decline 8 percent from 2023 to 2033 [32].

Industry Growth Numbers — Financial Compliance: Employment of compliance officers is projected to grow 5 percent from 2023 to 2033, with about 34,400 openings projected each year [42]. Employment of financial examiners is projected to grow 21 percent from 2023 to 2033, much faster than the average for all occupations [46].

AI Parallel Growth: This has created a surge in demand for compliance officers, risk managers, and legal advisors who can navigate regulatory complexities. Professionals with expertise in regulatory technology (RegTech) are particularly sought after [43]. 77% of new AI jobs require master’s degrees, creating substantial skills gaps [12].

Nuclear power plants employ sophisticated distributed control systems (DCS) and SCADA systems monitoring thousands of variables simultaneously, yet human operators remain essential for strategic oversight and emergency response. In finance, despite algorithmic trading’s dominance, regulatory requirements ensure human oversight remains mandatory for risk management, compliance, and strategic decision-making. These constraint-protected sectors demonstrate how certain contexts inherently require human judgment and accountability.

4. 🚀 The Medical Radiology Paradigm: Accountability, Agency and Edge-Cases

AI improves diagnosis 16%+, but doctors remain essential for clinical judgment

Medical radiology demonstrates how AI tools enhance rather than replace professional expertise, with significant performance improvements while maintaining human primacy.

Industry Growth Numbers: Overall employment of radiologic and MRI technologists is projected to grow 5 percent from 2024 to 2034, with about 15,400 openings projected each year [22]. The U.S. Bureau of Labor Statistics projects that overall employment of radiologic technologists is slated to grow 6% between 2022 and 2032, while MRI technologists will grow 8% between 2022 and 2032 [24]. In 2024, the average salary for a Radiologic Technologist in the United States was $86,484, a 12.3% increase from 2022 [30].

AI Parallel Growth: AI can make people more valuable, not less — even in the most highly automatable jobs. Revenue growth in AI-exposed industries has accelerated sharply since 2022 [13]. On average, AI was already either automating or augmenting some 25 percent of the day-to-day tasks across all jobs by the end of 2024 [20]. Healthcare-specific AI roles continue to emerge as hybrid positions combining human expertise with AI capabilities.

Medical radiology has refined computer-aided diagnosis (CAD) tools based on ML over the last two decades. Recent advances show remarkable results: ChatGPT improved chest X-ray diagnosis performance by 16.42 percentage points [13], and MOM-ClaSeg increased junior radiologists’ diagnostic sensitivity by 18.67% while reducing reading time by 27.07%.

Yet radiologists remain indispensable for interpreting AI findings within clinical context, making final diagnostic decisions, and communicating with healthcare providers. While simple tasks might be performed entirely by algorithms, the physician’s role in verifying outcomes, making clinical-epidemiological correlations, and determining treatment regimens remains irreplaceable [14]. This paradigm emphasizes human accountability, agency, and expertise in handling edge cases that require contextual judgment.

The Reality Gap: MIT’s Sobering Findings and Gartner’s Technology Maturity Insights

MIT’s AI Report 2025: The Implementation Reality Check

Recent research from MIT’s NANDA initiative provides crucial context to our analysis. MIT NANDA’s State of AI in Business 2025 reveals that after $30–40B of enterprise spend, roughly 95% of organizations are seeing no measurable P&L impact [1]. More significantly for our workforce analysis, disruption is currently only apparent in two industries — technology, and media and telecom [1].

This finding validates our supervision model paradigm. For most industries — professional services, healthcare and pharma, consumer and retail, financial services, advanced industries, and energy and materials — generative AI has had little impact beyond experimentation [1]. An unidentified COO at a mid-market manufacturing firm captured this reality: “The hype on LinkedIn says everything has changed, but in our operations, nothing fundamental has shifted. We’re processing some contracts faster, but that’s all that has changed” [1].

However, the MIT report reveals a critical insight: 90% of employees regularly use personal AI tools for work, even though only 40% of their companies have official AI subscriptions [1]. This “shadow AI economy” suggests that individual-level transformation is occurring even when enterprise-level change stalls [16].

In the affected sectors (Technology and Media), the impact is substantial: more than 80 percent of executives anticipate reduced hiring volumes within 24 months, while in healthcare and energy, “most executives report no current or anticipated hiring reductions over the next five years” [1].

Gartner’s Hype Cycle: Understanding AI Technology Maturity

Gartner’s 2025 Hype Cycle for Artificial Intelligence provides essential context for understanding which AI technologies are driving current job displacement and which represent future opportunities. The analysis shows a technology landscape in various stages of maturity [2]:

Peak of Inflated Expectations: AI agents and AI-ready data are the two biggest movers on this year’s Hype Cycle. Both sit at the Peak of Inflated Expectations [2]. This explains why many AI job displacement fears may be premature — the most talked-about autonomous AI systems are still experiencing “more hype than proof.”

Trough of Disillusionment: This year, GenAI enters the Trough of Disillusionment as organizations gain understanding of its potential and limits [2]. This shift explains why despite an average spend of $1.9 million on GenAI initiatives in 2024, less than 30% of AI leaders report their CEOs are happy with AI investment return [2].

Future Predictions: Gartner’s analysis suggests a measured but significant transformation ahead: at least 15% of day-to-day work decisions will be made autonomously through agentic AI by 2028, up from 0% in 2024, and 33% of enterprise software applications will include agentic AI by 2028, up from less than 1% in 2024 [2,3].

However, Gartner also provides a reality check on the agent hype: Over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls [2].

Lessons from the Supervision Models

These parallel industries reveal three consistent patterns that illuminate AI’s likely trajectory. Elevation, Not Elimination: Advanced automation doesn’t eliminate human work but elevates it to strategic oversight, creative problem-solving, and complex decision-making. Aviation pilots evolved from direct flight control to sophisticated system management, radiologists shifted from basic image reading to complex diagnostic interpretation, and nuclear operators moved from manual control to strategic oversight of automated systems. Automation relieves humans from repetitive tasks while invariably changing their active involvement into monitoring and exception-handling roles.

New Skill Requirements and Industry-Specific Adaptation: Success requires developing expertise in system monitoring, data interpretation, strategic thinking, and exception management, with 70% of job skills projected to change by 2030 and McKinsey estimating 30% of work hours could be automated within this decade [18]. Different sectors experience AI transformation at varying rates based on regulatory requirements, safety considerations, and human-centric service needs. The MIT findings validate this pattern — while Technology and Media sectors show substantial disruption with over 80% of executives anticipating reduced hiring within 24 months, healthcare and energy executives report “no current or anticipated hiring reductions over the next five years” [1]. This sector-specific adoption timeline, combined with Gartner’s analysis showing the most disruptive AI technologies still in early maturity stages, provides most workers time to develop supervision capabilities before their industry’s transformation accelerates.

Conclusion

The convergence of historical paradigms, MIT’s sector-specific research, and Gartner’s technology maturity analysis reveals that AI’s impact on work follows predictable patterns rather than unprecedented disruption. While 95% of organizations see no measurable P&L impact from $30–40B in AI spending, the 90% of employees already using AI tools personally suggests individual-level transformation is occurring even when enterprise adoption stalls [1]. The evidence supports a supervised intelligence model where humans evolve into strategic oversight roles, with 15–25% of jobs experiencing significant disruption by 2025–2027 concentrated initially in Technology and Media sectors [20,1]. The path forward requires sector-aware planning, personal AI tool adoption for practical experience, and developing strategic oversight capabilities that complement rather than compete with AI — joining the deliberate design of human-machine collaboration systems that aviation, radiology, and nuclear industries successfully implemented. Unlike previous technological revolutions, AI’s sector-specific adoption pattern gives most workers time to prepare, making the choice binary but not immediate: master the tools of supervision and strategic oversight, or risk being supervised by them when the technology matures.

References

[1] MIT Media Lab NANDA Initiative. (2025). The GenAI Divide: State of AI in Business 2025. Retrieved from https://nanda.media.mit.edu/ai_report_2025.pdf

[2] Gartner, Inc. (2025). Hype Cycle for Artificial Intelligence, 2025. Gartner Research.

[3] Gartner, Inc. (2025). The 2025 Hype Cycle for Generative AI Highlights Critical Innovations. Retrieved from https://www.gartner.com/en/articles/hype-cycle-for-genai

[4] PwC. (2025). The Fearless Future: 2025 Global AI Jobs Barometer. Retrieved from https://www.pwc.com/gx/en/issues/artificial-intelligence/ai-jobs-barometer.html

[5] National University. (2025). 59 AI Job Statistics: Future of U.S. Jobs. Retrieved from https://www.nu.edu/blog/ai-job-statistics/

[6] Nartey, J. (2025). AI Job Displacement Analysis (2025–2030). SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5316265

[12] National University. (2025). 59 AI Job Statistics: Future of U.S. Jobs. Retrieved from https://www.nu.edu/blog/ai-job-statistics/

[13] PwC. (2025). The Fearless Future: 2025 Global AI Jobs Barometer. Retrieved from https://www.pwc.com/gx/en/issues/artificial-intelligence/ai-jobs-barometer.html

[14] Reith, W., et al. (2023). Redefining Radiology: A Review of Artificial Intelligence Integration in Medical Imaging. Diagnostics, 13(17), 2760. https://doi.org/10.3390/diagnostics13172760

[15] McKinsey & Company. (2025). AI in the workplace: A report for 2025. Retrieved from https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work

[16] Masood, A. (2025). The GenAI Divide: MIT NANDA’s research on what’s real, what’s working, and what leaders should do next. Medium. Retrieved from https://medium.com/@adnanmasood/the-genai-divide-mit-nandas-research-on-what-s-real-what-s-working-and-what-leaders-should-do-26a9fe53e0b4

[17] AIPRM. (2024). 50+ AI Replacing Jobs Statistics 2024. Retrieved from https://www.aiprm.com/ai-replacing-jobs-statistics/

[18] Nexford University. (2025). How will Artificial Intelligence Affect Jobs 2025–2030. Retrieved from https://www.nexford.edu/insights/how-will-ai-affect-jobs

[20] AIMUltiple. (2025). Top 17 Predictions from Experts on AI Job Loss. Retrieved from https://research.aimultiple.com/ai-job-loss/

[21] AI Statistics 2024–2025: Global Trends | Founders Forum Group. Retrieved from https://ff.co/ai-statistics-trends-global-market/

[22] U.S. Bureau of Labor Statistics. Radiologic and MRI Technologists: Occupational Outlook Handbook. Retrieved from https://www.bls.gov/ooh/healthcare/radiologic-technologists.htm

[24] Barton Healthcare Staffing. (2024). 2024 Medical Imaging Job Outlook By Specialty. Retrieved from https://www.bartonhealthcarestaffing.com/blog/2024-medical-imaging-job-outlook-by-specialty/

[30] RSG Health Services. (2025). 2025 Medical Imaging Compensation Review. Retrieved from https://radsciences.com/2025-medical-imaging-compensation-jobs-review/

[32] U.S. Bureau of Labor Statistics. Power Plant Operators, Distributors, and Dispatchers: Occupational Outlook Handbook. Retrieved from https://www.bls.gov/ooh/production/power-plant-operators-distributors-and-dispatchers.htm

[41] World Nuclear News. (2024). US nuclear workforce continues to grow, report finds. Retrieved from https://www.world-nuclear-news.org/Articles/US-nuclear-workforce-continues-to-grow,-report-fin

[42] U.S. Bureau of Labor Statistics. Compliance Officers: Occupational Outlook Handbook. Retrieved from https://www.bls.gov/ooh/business-and-financial/compliance-officers.htm

[43] STC. (2025). Finance Jobs Outlook — Mid-Year 2025. Retrieved from https://www.stcusa.com/resource-center/career/finance-jobs-outlook/

[46] U.S. Bureau of Labor Statistics. Financial Examiners: Occupational Outlook Handbook. Retrieved from https://www.bls.gov/ooh/business-and-financial/financial-examiners.htm

[51] Wikipedia. (2025). Dot-com bubble. Retrieved from https://en.wikipedia.org/wiki/Dot-com_bubble

[52] Sanyal, S. (2024). Dot-com bubble vs AI boom: Lessons for today’s market. LinkedIn. Retrieved from https://www.linkedin.com/pulse/dot-com-bubble-vs-ai-boom-lessons-todays-market-sugata-sanyal-crerc

[53] Bank for International Settlements. (2024). Is today’s AI boom bigger than dotcom bubble? BIS Quarterly Review. Retrieved from https://www.bis.org/publ/qtrpdf/r_qt2409w.htm

[54] Reuters. (2025). Is today’s AI boom bigger than dotcom bubble? Retrieved from https://www.reuters.com/markets/europe/is-todays-ai-boom-bigger-than-dotcom-bubble-2025-07-22/

[55] Intel Ignite. (2024). Lessons from dot-com era. Retrieved from https://intelignite.com/lessons-from-dot-com-era/

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI