AI and work
How AI began changing knowledge and creative work — from Copilot and ChatGPT to the WGA and SAG-AFTRA contracts and the first evidence of AI adoption depressing entry-level hiring.
This thread follows AI’s contact with labour — what it does to the work people are paid for. It began with the tools: GitHub Copilot for programmers, and then ChatGPT, which put a capable writing and analysis assistant in front of every knowledge worker at once.
Organised labour responded before the economists did. Hollywood’s 2023 disputes produced the first contracts to constrain the technology: the WGA’s screenwriting guardrails and SAG-AFTRA’s consent rules for digital replicas set terms for how studios could use AI on writers’ and performers’ work.
The tools then grew more capable of doing the work rather than assisting it. Claude 3.5 Sonnet and Artifacts changed how people used a chatbot; computer use and agents like Deep Research took on multi-step tasks end to end. The first hard labour-market evidence arrived in 2025, when a Stanford study linked AI adoption to falling employment for young workers in the most exposed occupations. Whether AI augments workers or displaces them is the thread’s central question, and that early signal pointing at entry-level jobs is the first real evidence either way.
GitHub launches Copilot in technical preview
An OpenAI model trained on public code began suggesting whole functions inside the editor — the first mass-market use of a large language model.
Models & capabilities · Culture & impact
OpenAI publishes Codex and the HumanEval benchmark
Codex solved 28.8% of HumanEval's Python problems on a single attempt and 70.2% when allowed 100 samples per problem, against 0% for base GPT-3.
Models & capabilities · Benchmarks & progress
DeepMind's AlphaCode reaches median human on competitive programming
Ranked around the 54th percentile in Codeforces contests by generating and filtering enormous numbers of candidate programs.
Models & capabilities · Benchmarks & progress
GitHub Copilot goes on sale
Priced at $10 a month or $100 a year, it moved from a year-long technical preview used by over 1.2 million developers to a paid product free for students and open-source maintainers.
Money & business · Models & capabilities
An AI image wins the Colorado State Fair art prize
Jason Allen's Midjourney piece 'Théâtre D'opéra Spatial' took first place in the fair's digital arts category and a $300 prize, and was later denied US copyright registration.
Culture & impact
OpenAI launches ChatGPT
A free web interface to an existing GPT-3.5 model, reported by one analysis to have reached 100 million monthly users within two months.
Models & capabilities · Culture & impact
Stack Overflow bans ChatGPT answers
Within a week of launch, a major knowledge community prohibited model output because plausible wrong answers arrived faster than moderators could check them.
Culture & impact
New York City schools block ChatGPT
The city's Department of Education cited weak critical-thinking value and cheating risk; it reversed the block on ChatGPT four months later.
Culture & impact
Microsoft announces 365 Copilot
Built on GPT-4 and Microsoft's own data graph, it launched to a small group of testers with no price or general release date set.
Money & business · Culture & impact
BuzzFeed News shuts down as parent company pivots to AI content
About 180 staff, 15% of BuzzFeed Inc, were cut; the company said no journalists were being replaced by AI even as it kept expanding AI-generated content elsewhere.
Culture & impact · Money & business
IBM pauses hiring for roles it expects AI to replace
The estimate covered non-customer-facing roles over five years; IBM later said the actual toll was a couple of hundred HR positions, offset by hiring elsewhere.
Culture & impact · Money & business
Chegg stock falls 48% as CEO says ChatGPT is hurting new customer growth
Rosensweig called the plunge 'extraordinarily overblown' a day later, but by late 2025 the company had laid off roughly 45% of its remaining staff.
Culture & impact · Money & business
Meta releases Code Llama
Released in four sizes up to 70B parameters under Llama 2's licence, the largest variant reportedly matched ChatGPT on the HumanEval coding benchmark.
Open weights & ecosystem
WGA ratifies contract with new guardrails on AI in screenwriting
Members approved the deal 99% to 1%, on 8,525 votes cast, five months after AI protections became a central demand of the strike that began in May.
Culture & impact
SAG-AFTRA's 2023 film/TV contract sets consent rules for digital replicas
The deal required separate, informed consent and compensation before a studio could create or reuse a performer's AI-generated digital double.
Culture & impact
Klarna says AI does the work of 700 customer service agents
The buy-now-pay-later firm said its assistant handled 2.3 million chats in a month, cut resolution time from 11 minutes to under two, and projected $40m in 2024 profit gains.
Culture & impact · Money & business
Cognition demos Devin, billed as the first AI software engineer
Cognition said Devin resolved 13.86% of real GitHub issues unassisted on the SWE-bench benchmark, against roughly 2% for the prior best system.
Models & capabilities · Culture & impact
UK Society of Authors survey: a third of translators losing work to generative AI
787 members surveyed; 36% of translators and 26% of illustrators reported lost work, while over a third of translators had already used generative AI themselves.
Culture & impact
Claude 3.5 Sonnet and Artifacts change how people use chatbots
Priced and sped like Anthropic's mid-tier model, it scored 64% on the company's internal agentic-coding evaluation against 38% for the outgoing flagship.
Models & capabilities
YouGov: nearly half of employed Americans expect AI to shrink jobs in their industry
A YouGov survey found 48% of employed Americans expected AI to reduce jobs in their industry, up from 29% in March 2023, though only 2% reported it had already cost them work.
Culture & impact
Claude gets computer use
The public beta let Claude view screenshots and issue cursor, click and keystroke commands, scoring 14.9% on OSWorld against 7.8% for the nearest rival.
Models & capabilities
Coca-Cola's AI-generated Christmas ad draws backlash over 'soulless' imagery
Coca-Cola's AI-generated holiday ad, remaking its classic 1995 truck commercial, was widely mocked online as uncanny and 'soulless', becoming a flashpoint in advertising's AI debate.
Culture & impact
Amazon expands AI audiobook narration, drawing voice-actor and author criticism
Audible's AI narration catalogue passed 50,000 titles and the company added translation and 100-plus synthetic voices, drawing objections from narrators' unions.
Culture & impact
OpenAI ships Deep Research
An agent that browsed for tens of minutes and returned cited reports, the first widely used long-horizon research tool.
Models & capabilities
Anthropic launches the Anthropic Economic Index
Analysis of roughly one million anonymised Claude.ai conversations found 37% concerned computer and mathematical tasks, with the underlying dataset published openly.
Culture & impact
Anthropic ships Claude 3.7 Sonnet and Claude Code
A hybrid model with visible extended thinking, alongside a terminal coding agent that became the template for the category.
Models & capabilities
Danish study finds AI chatbots barely moved wages or hours despite time savings
Survey and payroll data on 25,000 Danish workers across 11 occupations found chatbots saved about 3% of work time but left earnings and hours unchanged two years on.
Culture & impact
Duolingo's 'AI-first' memo triggers user backlash and app-deletion campaign
CEO Luis von Ahn said Duolingo would phase out contractors AI could replace and weigh staff AI use in reviews; he later said the memo lacked context.
Culture & impact
Business Insider cuts 21% of staff in 'all-in on AI' pivot
Its third round of layoffs in three years, announced the same week the company hired a newsroom AI lead; the union called it a pivot 'toward greed.'
Culture & impact · Money & business
Study finds AI wrote about 30% of US developers' Python code by late 2024
A classifier trained on 31 million GitHub commits found gains concentrated among experienced developers, while beginners barely benefited despite adopting the tools at similar rates.
Ideas & essays · Culture & impact
SAG-AFTRA video game performers suspend strike after AI deal
The strike had run since July 2024; performers can now suspend consent for AI digital replicas of themselves during any future strike.
Culture & impact
Vox Media union ratifies contract with AI job-security and disclosure protections
The roughly 250-member unit had voted 90% to authorise a strike before management agreed, on the eve of the old contract's expiry, to protections against AI replacement.
Culture & impact
MIT report finds 95% of generative AI enterprise pilots show no measurable return
MIT's NANDA project reviewed over 300 public AI deployments and interviewed enterprise leaders, finding most generic pilots stalled while custom, workflow-specific tools succeeded.
Money & business · Culture & impact
Stanford study finds AI adoption linked to falling employment for young workers
Software developers aged 22 to 25 saw the sharpest declines; the authors found no comparable fall in wages, meaning employers cut headcount rather than pay.
Culture & impact
Stanford/BetterUp study puts a price on AI 'workslop' in the workplace
Roughly half of workers who received the low-effort AI output said it made them see the sending colleague as less capable, creative or trustworthy.
Culture & impact
OpenAI publishes GDPval, a benchmark for economically valuable knowledge work
Blind grading by industry professionals rated GPT-5 and Claude Opus 4.1 outputs as equal to or better than human work on nearly half of the 1,320 tasks.
Benchmarks & progress
Anthropic reports early estimates of Claude's productivity gains
Analysing 100,000 Claude.ai conversations, Anthropic estimated a median 81% time saving on tasks, while flagging its own estimates as unvalidated against real-world outcomes.
Money & business
Pew: two-thirds of US teens now use AI chatbots
ChatGPT was used by 59% of teen chatbot users, more than double Gemini or Meta AI, and one in ten said they relied on a chatbot for all or most of their schoolwork.
Culture & impact
Paper projects AI could raise US productivity ~20% over a decade
Merali split the productivity gain roughly 56% compute scaling and 44% algorithmic progress, and found it far smaller for agentic tasks needing tool use.
Ideas & essays
Anthropic study finds heavy AI-coding use can reduce skill formation in junior engineers
In a randomised trial of 52 mostly-junior engineers learning a new library, the hand-coding group scored 67% on a comprehension quiz against 50% for those given AI assistance.
Culture & impact
Anthropic Economic Index reports on how Claude use changes as users gain experience
Analysing a million Claude conversations from one week in February 2026, Anthropic found six-month-plus users had roughly 10% higher task success rates than newcomers.
Culture & impact
Harvey raises $200M at $11B valuation
The $11B figure is up from an $8B valuation just three months earlier, in a round co-led by GIC and Sequoia, which had now backed Harvey three times.
Money & business
OpenAI proposes 'Industrial Policy for the Intelligence Age'
The 13-page document floats a tax on automated labour, a citizen wealth fund modelled on Alaska's, and government-backed 32-hour-week pilots, while calling itself a starting point rather than a recommendation.
Government & policy
METR survey finds software engineers reporting ~2x AI speedup
The 349-respondent convenience sample also reported a 3x median speed gain, but METR flagged that self-reported estimates have previously overstated AI's effect by 40 percentage points against controlled measurement.
Benchmarks & progress · Culture & impact
OpenAI launches its own deployment/compute company
Despite the name, it is not a compute or infrastructure venture but a consulting-style business embedding OpenAI engineers in client organisations, launched with over $4 billion from 19 outside investors.
Money & business · Labs & people
California governor signs executive order on AI and workforce displacement
The order gives state agencies 180 days to review unemployment-insurance rules and directs new dashboards tracking AI's effect on jobs, without creating enforceable obligations on employers.
Government & policy · Culture & impact
UC Berkeley releases Agents' Last Exam, a benchmark of professional work
Built with 250+ industry experts across 55 sub-industries, it runs agents in the real software a specialist would use and grades against hidden answers; current systems clear under 1% of the hardest tier.
Benchmarks & progress
AI cited as leading cause of US layoffs for first time, Challenger report finds
Employers cited AI as the reason for 40% of May's 97,006 announced job cuts, up from 7% in January, with the technology sector accounting for the largest share.
Culture & impact
Musicians' union sues Universal and Warner over their Suno and Udio AI licensing deals
Filed in the Southern District of New York, the suit argues session musicians are owed a share of settlement and licensing revenue under a 'new use' contract clause.
Courts & copyright · Money & business
Anthropic publishes Economic Index report on usage 'cadences'
Anthropic found Claude usage tracks daily and weekly rhythms — a 2.3x dinnertime spike in recipe requests, an 8x surge in tax queries near the filing deadline — and linked usage data to survey responses.
Culture & impact
Gallup finds Americans growing more negative on AI as familiarity rises
Gallup found 39% of Americans said AI does more harm than good, up from 31% a year earlier, nearing the 40% recorded in 2023 after two years of improvement.
Culture & impact