Someone at the event told me they forwarded AI-generated information that turned out wrong. After that, they stopped trusting the tool entirely. One bad output. That was it.
That story stayed with me more than any of the talks.
I spent Thursday at the Institution of Civil Engineers for a Rail in AI event. Four speakers. Good turnout. Honest questions. The jobs question came up in every Q&A. It is a fair concern. I have trusted AI to do work it should not have done. I asked it about specific domain knowledge it simply did not have the training for.
But I have also been paying attention to the data coming out this year. The studies are better than they were. Real spending data. Actual headcount records. Not surveys.
I checked three studies.
Three studies
Ramp and Revelio Labs published a paper in June 2026 that links actual AI vendor payments to employment records across 21,559 US firms. Not self-reported survey data. Real money flowing to OpenAI, Anthropic, and other providers, matched to payroll records.
High-intensity adopters grew headcount 10.2% over two years. Entry-level roles grew 12%. The effect was broad: engineering, sales, admin, customer service, finance. Not just technical roles.
The study has caveats. Adopters were already larger and faster-growing. The clearest gains show up in the Information sector. But it is the best measure I have seen of what happens when companies actually invest in AI rather than experiment with it.
The PwC 2026 Global AI Jobs Barometer covers over a billion job ads across six continents. The most AI-exposed companies see faster headcount growth than the least exposed: 52% versus 36%. They also see higher wage growth: 24% versus 17%. Productivity growth is 40% higher in the most exposed companies.
The Barometer does not claim AI causes growth. It documents a correlation. Firms that adopt AI deeply tend to hire more, not less.
MIT's Project NANDA published The GenAI Divide in July 2025. It is the sobering one. Despite $30-40 billion in enterprise generative AI investment, about 95% of organisations reported no measurable P&L impact from their AI pilots.
The 95% figure got a lot of attention. The more useful finding: pilots built with vendors or partners succeeded 67% of the time. Pure internal builds succeeded one-third as often.
What the UK data says
The US numbers are useful, but I work in the UK. So I checked what is happening here.
Employment Hero surveyed 3,500 UK employers in 2026. Businesses with AI at the core of their operations were more than twice as likely to have increased entry-level headcount over the past two years: 62% versus 30% for non-adopters.
The UK is leading on this measure. 24% of UK business leaders expect AI to increase the need for entry-level roles. That is nearly double the rate in Australia (13%) or Canada (15%).
The ONS published its own data in July 2026 from the Business Insights and Conditions Survey, covering 38,637 responding businesses. Around 35% of UK businesses with 10 or more employees now use at least one AI technology, up from 12% in late 2023.
Most of those businesses report no change to headcount so far. Where decreases show up, they are concentrated in medium-sized firms (100-249 employees), where just under 7% reported a decrease.
The British Chambers of Commerce found 54% of UK firms actively using AI in March 2026. 95% of SMEs using AI said it had no impact on workforce size. But a smaller group adopting deeper, bespoke AI were more likely to expect headcount reductions.
The story is consistent: companies that adopt AI seriously hire more. Companies that dabble see no change. The UK data is less dramatic than the US study, but it points the same direction.
The gap at the event
One speaker, Jiaxi Li, talked about operational research in UK rail. The point that stayed with me: UK rail underinvests in OR compared to aviation and logistics.
Aviation has used AI crew scheduling for years. Route optimisation at UPS saves $300-400 million a year. DHL's robot-assisted sorting increased capacity by 40%. UK rail has network modelling tools but the investment and data quality lag behind.
That is not an AI problem. It is a prioritisation problem.
One thing that was not discussed
Anthropic confidentially filed for IPO with the SEC on June 1. Its Series H funding round valued the company at $965 billion. Annualised revenue run-rate: $47 billion, up from $9 billion at the end of 2025.
A valuation of that size depends on a narrative. AI as a force reshaping industries supports it. AI as a tool that helps engineers draft documents faster does not.
I am not saying the jobs concern is manufactured. It is real. But the incentive to amplify the danger narrative is also real.
What I think
The data is consistent: serious AI adoption correlates with employment growth. The companies that invest deeply grow headcount faster. The projects that fail do so because of organisational problems, not technical ones.
But data describes averages. It does not describe your project or your railway.
The practical question from the event was direct: how did you do your timesheet last month? That is the level to start at. Find the repetitive work. Automate that. Prove the methodology. Scale.
Senior engineers can spot when AI is wrong. Juniors cannot. That is the real risk. Not replacement. A generation of engineers who do not build the judgement to challenge an AI output because it did the work before they had to.
The solution is not stopping AI. It is good leadership, structured mentorship, and letting juniors make mistakes in environments where the consequences are contained.
The gap in rail is not an imagination problem. It is an adoption and trust problem. The sector knows what AI could do. It has not built the confidence or the skills to do it well.
That is the work. Not the technology.
Sources
- Ramp Economics Lab / Revelio Labs: "A New Look at AI's Impact on Jobs: Firm-Level AI Spending and Workforce Adjustment". Kharazian, Simon & Stevens (June 2026). ramp.com
- PwC 2026 Global AI Jobs Barometer: Global findings (2026). pwc.com
- MIT Project NANDA: "The GenAI Divide: State of AI in Business 2025" (July 2025). valtao.com
- Employment Hero / Focaldata: "The AI Paradox at Work" (2026). 3,500 UK employers surveyed. staffingindustry.com
- ONS: Business Insights and Conditions Survey, Wave 159 (July 2026). 38,637 responding UK businesses.
- British Chambers of Commerce / ISER: "Future of Work: AI in the Workplace Report" (March 2026). iser.essex.ac.uk
- Fortune: "Anthropic confidentially files for IPO after a $965 billion valuation" (June 1, 2026). fortune.com
- The Wall Street Journal: "Anthropic Files to Go Public in Blockbuster Year for IPOs" (June 1, 2026). wsj.com
- Research and Markets: Railway AI market report (February 2026). researchandmarkets.com
- UPS ORION: Route optimisation savings. UPS corporate communications.
- techUK: UK rail data fragmentation and innovation barriers.
Image credits
- Header image: Institution of Civil Engineers, One Great George Street. CC BY-SA 2.0, by Derek Harper, via Geograph / Wikimedia Commons. Source file
