Former Tory chancellor tasked with helping ChatGPT owner develop ties with governments
The former UK chancellor George Osborne is joining OpenAI to lead the ChatGPT developer’s relationships with governments around the world.
He will head a division known internally as OpenAI for Countries, through which the San Francisco artificial intelligence startup works with governments on national-level AI rollouts.
McKinsey, the consulting giant that has spent a century advising companies on how to cut costs and restructure operations, is now turning that advice inward as it plans to eliminate thousands of jobs across its non-client-facing departments over the next 18 to 24 months.
The firm's leadership has discussed a roughly 10% headcount reduction in support functions, according to Bloomberg. McKinsey's revenue has hovered around $15 billion to $16 billion for the past five years after a decade of rapid expansion that saw employee count climb from 17,000 in 2012 to 45,000 by 2022. The headcount has since slid to about 40,000.
The cuts come as consulting firms face cost-conscious clients, Trump administration pressure on government consulting spending, and reduced payments from Saudi Arabia, which had been paying McKinsey at least $500 million annually in the decade up to 2024. McKinsey cut about 1,400 jobs in 2023 under a plan internally labeled Project Magnolia, and axed 200 global tech positions last month. The firm still plans to hire consultants even as it shrinks support staff.
Producer Stephen Woolley pays tribute to Figgis, who has died aged 77, a brilliant professional whose ‘molotov cocktail personality’ enabled her work in British and Hollywood cinema
I first encountered Susie Figgis over 40 years ago when I interviewed her for The Company of Wolves, my debut movie production with Neil Jordan. We met at my then-cinema the Scala – it was a busy, noisy office but a sunny day, so we went up to the roof. Susie, who was already something of a legend having cast Stephen Frears’ Bloody Kids, Laura Mulvey’s avant garde films and Ben Kingsley in Gandhi, unleashed a volcanic eruption of unbridled enthusiasm for Angela Carter and Neil’s script. The collection of explosive expletives and voluble “darlings” almost blasted me to the King’s Cross streets below.
So began a professional relationship that spanned more than 23 movies. The task we set her for The Company of Wolves was tricky: to find an actor to play the adolescent Rosaleen. She achieved it through painstaking and meticulous methods (her trademark) over the next few months, exceeding our expectations when she discovered the excellent Sarah Patterson. She then topped that with the suggestion of Angela Lansbury for “Grannie” (who flew from Hollywood to shoot with us and had her character’s head decapitated for her troubles) and a superlative supporting cast of dancers, performance artists and veteran actors for our strange, violent woodland fairytale. Her passion for cinema was infectious: not only for the film-makers, but also the agents and actors who read our scripts. Susie demanded an intelligent and thoughtful response to the screenplays so no simple yes or no would suffice.
Fewer companies operating in Europe will be made to carry out due diligence on the societal harms they cause, in what green groups have called a “betrayal” of communities affected by corporate abuse.
The gutting of the EU’s sustainability reporting and due diligence rules, which was greenlit by MEPs on Tuesday, slashes the number of companies covered by laws to protect human and ecological rights, and removes provisions to harmonise access to justice across member states.
Equinor accused of ‘extensive and long-term pollution’ caused by years of inadequate maintenance
Norway’s national oil company, Equinor, is facing a £53m penalty for oil spills and gas leaks at the oil-rich Scandinavian state’s only refinery, which officials said were the result of years of inadequate maintenance.
Norway’s economic crime agency, Økokrim, said it had taken action against Equinor over “extensive and long-term pollution” at the refinery in Mongstad, on Norway’s North Sea coast.
Data shows the headline unemployment rate continued to climb and hit 4.6%, a four-year high, last month
The US labor market grew by more than expected last month, recovering some of the damage inflicted by the federal government shutdown, according to official data.
An estimated 105,000 jobs were lost in October, and 64,000 were added in November, a highly-anticipated report showed on Tuesday.
Heavily indebted utility puts back ‘retention payments’ for 21 executives until new year amid search for rescue deal
Thames Water has deferred awarding bosses retention payments totalling £2.5m, avoiding a potentially damaging pre-Christmas row as the heavily indebted utility scrambles to agree a multibillion-pound rescue deal.
Sources at the UK’s biggest water company confirmed the controversial retention payment package for 21 senior executives, which had been due to go out this month, would remain on hold until the new year.
When Labour came to power, it set a “long-term ambition” of increasing the employment rate – the share of the working age population with a job – to 80%. The latest data suggest things are moving in the wrong direction.
The employment rate in the three months to October was 74.9%, according to the Office for National Statistics (ONS), down 0.3 percentage points on the quarter.
The UK supermarket chain Morrisons faces a £17m tax bill after losing a lengthy court battle against HMRC over the charging of value added tax (VAT) on rotisserie chicken.
The high court has ruled that whole cooked cool-down chickens should be subject to the standard 20% VAT rate for hot food.
Rolling coverage of the latest economic and financial news
ING’s UK economist JamesSmith has spotted that government hiring is no longer supporting the jobs market.
He writes:
Companies – especially in retail and hospitality – have been shedding workers this year, partly because of earlier tax and minimum wage hikes. Hiring surveys remain weak.
Until recently, that was helpfully offset by resilience in government hiring, but that appears to be changing. Public sector employment has also now fallen for three consecutive months, judging by those payroll numbers.
Companies such as Udio, Suno and Klay will let you use AI to make new music based on existing artists’ work. It could mean more royalties – but many are worried
This was the year that AI-generated music went from jokey curiosity to mainstream force. Velvet Sundown, a wholly AI act, generated millions of streams; AI-created tracks topped Spotify’s viral chart and one of the US Billboard country charts; AI “artist” Xania Monet “signed” a record deal. BBC Introducing is usually a platform for flesh-and-blood artists trying to make it big, but an AI-generated song by Papi Lamour was recently played on the West Midlands show. And jumping up the UK Top 20 this month is I Run, a track by dance act Haven, who have been accused of using AI to imitate British vocalist Jorja Smith (Haven claim they simply asked the AI for “soulful vocal samples”, and did not respond to an earlier request to comment).
The worry is that AI will eventually absorb all creative works in history and spew out endless slop that will replace human-made art and drive artists into penury. Those worries are being deepened by how the major labels, once fearful of the technology, are now embracing it – and heralding a future in which ordinary listeners have a hand in co-creating music with their favourite musicians.
Chancellor’s claim to be helping trade met with disbelief in England and Wales amid soaring staff costs, energy bills and other overheads
Emma Harrison has begun to wonder how her business will survive in recent weeks. The managing director of the Three Hills pub in Bartlow, Cambridgeshire, is struggling to see how she will make a profit after examining the impact of her rising tax bill.
“I’m really terrified about this coming year,” Harrison says. “We’re a well-run pub, we’ve won lots of awards, but this is going to be really hard.”
Demands for oversight grow after inquiry calls sector an ‘unregulated free for all’ and families seek stronger safeguards
Ministers are expected to back calls to regulate England’s funeral industry for the first time, after a series of scandals over the handling of remains.
Bereaved families have called for a new investigatory body and rules governing professional qualifications after an official inquiry declared the sector an “unregulated free for all”.
Ten years after I first followed the proposed route, I retraced my steps to see what life was like along the world’s most expensive, heavily delayed railway line
Ten years ago, I walked the route of HS2, the 140-mile railway proposed to run from London to Birmingham, to discover what lay in its path. Nothing had actually been constructed of this, supposedly the first phase of a high-speed line going north. The only trace was the furtive ecological consultants mapping newts and bats and the train’s looming presence in the minds of those who lived along the route. For many, it was a Westminster vanity project, symbolising a country run against the interests of the many to line the pockets of the few. People whose homes were under threat of demolitionwere petitioning parliament, campaigning for more tunnels or hoping the project would collapse before their farms, paddocks and ancient woodlands were wiped out.
The line, we were told a decade ago, would be completed by 2026. Like many of the early claims about the longest railway to be built in Britain since the Victorian era, that fact no longer stands. The fast train is running – very – late. The official finish date of 2033 was recently revised upwards. “The best guess is that it will begin with a ‘4’ when you can catch a train,” one well-informed observer told me. There’s similar uncertainty about its cost, but one thing is sure: it is catastrophically over budget. When complete, HS2 will almost certainly be the most expensive railway in the world. Nearly 20 years ago, HS1, the line from the Channel tunnel to St Pancras, was completed on time and on budget for £51m per mile (£87m in today’s prices). It was criticised for being twice as expensive as a high-speed route constructed in France. HS2 may cost almost £1bn per mile.
First new treatments for sexually transmitted disease in decades approved by US Food and Drug Administration as number of cases worldwide surge to 82m
The first new treatments for gonorrhoea in decades could be a “huge turning point” in efforts to combat the rise of superbug strains of the bacteria, researchers have said.
Gonorrhoea is on the rise around the world, with more than 82m infections globally each year and particularly high rates in Africa and countries in the World Health Organization’s Western Pacific region, which reaches from Mongolia and China to New Zealand. Cases in England are at a record high, and rates in Europe were three times higher in 2023 than in 2014.
Launch of third generation of Leaf follows investment of more than £450m into model, including £300m direct to UK
Nissan has started the production of its latest electric car in Sunderland, a crucial step in the UK automotive industry’s transition away from petrol and diesel.
The Japanese manufacturer will launch the third generation of the Leaf on Tuesday, which was the first mass-market battery electric car to be built in the UK. Nissan has made 282,704 Leaf models at the north-east England plant so far.
Once a star of the self-driving hype cycle, lidar maker Luminar has filed for bankruptcy amid legal turmoil, layoffs, and a cooling autonomous-vehicle market. It plans to sell off its assets before shutting down entirely. The Verge reports: As part of its bankruptcy, Luminar is seeking permission to sell both its lidar and semiconductor businesses, the latter of which it has already agreed to sell to Quantum Computing for $110 million. The company plans to continue to operate during the bankruptcy proceedings "to minimize disruptions and maintain delivery of its LiDAR hardware and software." That said, Luminar will cease to exist once the process is complete. "As we navigate this process, our top priority is to continue delivering the same quality, reliability and service our customers have come to expect from us," CEO Paul Ricci said in a statement.
After launching in 2017, Luminar muscled its way to the front of the autonomous vehicle industry as a top maker of lidar systems, a key technology that driverless cars use to sense the shapes and distances of objects around them. Luminar has sold sensors to Mercedes-Benz, Volvo, Audi, Toyota Research Institute, Caterpillar, and even Tesla, which has dismissed lidar sensors in favor of traditional cameras. The company was valued at nearly $3 billion when it went public through a reverse merger with a SPAC in 2020.
Government says arrangement will bring in extra £400m on top of more than £15bn of existing annual trade with Korea
The UK has signed a new trade deal with South Korea designed to increase exports of cars, Scottish salmon and Guinness canned in Britain.
Keir Starmer described the deal, which replaces an existing agreement, as “a huge win for British business and working people”. It follows UK deals with India and the US, and the free trade agreement with the EU clinched this year.
Company to scrap several electric models and focus on gas and hybrid as US president pulls support for EVs
Ford said on Monday it will take a $19.5bn writedown and is killing several electric-vehicle models, in the most dramatic example yet of the auto industry’s retreat from battery-powered models in response to the Trump administration’s policies and weakening EV demand.
Ford, based in Dearborn, Michigan, said it will stop making the F-150 Lightning in its electric vehicle form, but will pivot to producing an extended-range electric model, a version of a hybrid vehicle called an Erev, which uses a gas-powered generator to recharge the battery.
Pledge to invest billions in UK paused, with Washington citing lack of progress on trade barriers across pond
The US has paused its promised multi-billion-pound investment into British tech over trade disagreements, marking a serious setback in US-UK relations.
The £31bn “tech prosperity deal”, hailed by Keir Starmer as “a generational stepchange in our relationship with the US” when it was announced during Donald Trump’s state visit, has been put on ice by Washington.
Britain’s fifth-biggest grocer postpones goal by 15 years to 2050, saying revised plan will now cover entire supply chain
Morrisons has become the first UK supermarket chain to postpone its net zero carbon emission targets, delaying them by 15 years to 2050.
Britain’s fifth-biggest grocer said its new targets would cover the entire supply chain, as well as Morrisons stores, including emissions from agriculture and land-use sources.
Britain’s biggest business groups have urged Conservative peers to stop blocking Labour’s workers’ rights bill in the House of Lords to avoid throwing away a compromise deal reached with trade unions.
With the clock ticking before Christmas, six of the country’s biggest employers’ groups warned that failure to pass the legislation before parliament rises on Thursday could put at risk a deal brokered with bosses and union leaders.
Will the employment rights bill be passed by Christmas? Well, the chances are slightly improved after six leading business groups published a temperature-lowering letter on Monday that said parliament, which in this instance means the blockers in the House of Lords, should get on with it.
The employers, note, are still unhappy about the issue that triggered the most recent revolt by Conservative peers and a few cross-benchers: the removal of a cap on compensation claims for unfair dismissal. But they’re more worried that further delays would jeopardise their negotiating victory last month, namely the government’s U-turn on rights guaranteeing workers protection against unfair dismissal from day one of employment. A six-month qualifying period was adopted instead, with the blessing of the TUC, which was similarly motivated by trying to get the bill over the line quickly.
Reflecting on chaos of early pandemic, former chancellor said it was ‘acutely stressful’ to see rising interest bill on government bonds
Rishi Sunak was concerned about the UK’s ability to fund itself in March 2020 after the government announced rescue measures costing tens of billions of pounds to prevent mass redundancies, the Covid-19 pandemic inquiry has heard.
The former prime minister, who was chancellor when the first UK lockdown was announced, said he feared foreign investors had become more concerned about Britain’s ability to pay its way than other countries in a similar situation.
Some potential first-time buyer groups ‘could be better served’, regulator says in review
Freelancers and gig economy workers could enjoy more flexibility over how and when they pay their mortgage under plans designed to help more people get on the property ladder.
A shake-up of the rules so people whose income is “variable or irregular” could be freed up from having to make monthly mortgage payments is one of a number of changes being considered by the Financial Conduct Authority (FCA) that could make it easier for millions of “underserved” UK consumers to get a home loan.
City’s small shops are reaching tipping point amid higher business rates, staff costs and big chains eager to move in
It’s lunchtime at Dormitory, an independent bedlinen store on Gloucester Road in Brighton, and proprietors Sue Graham and Cathy Marriott are peering across the street at the Brighton Sausage Co. They can tell when shoppers have stayed indoors by the number of sausage rolls left in the window. It’s a Tuesday before Christmas – supposedly the busiest time of the year. But there’s still a big pile remaining.
“In 10 years’ time, we’re all going to be going, ‘We need shops. Where have they all gone?’,” Marriott says. Her warning echoes widespread fears for Brighton’s plentiful independent shops, which have given the Sussex city international renown.
The chances of the European trucking industry hitting zero emissions targets are “dire”, an industry body has warned, as it emerged that only a tiny amount of lorries delivering goods in the EU are electric.
Speaking as the European Commission prepares to water down electric car targets, the boss of the association for commercial vehicles called on the commission to commit to an urgent review of the sector, tackling problems including a lack of public charging points, a lack of tax breaks for trucks and high energy costs.
AA sounds out buyers, and owners of fellow roadside assistance provider prefer IPO of similar value
The private equity owners of the AA, Britain’s biggest roadside recovery business, are looking for a potential £5bn sale or stock market flotation, while the owners of the rival RAC are targeting a London listing with a similar valuation.
The AA, which provides roadside assistance as well as insurance and driving lessons, is owned by a consortium including TowerBrook Capital Partners, Warburg Pincus and Stonepeak, and has been sounding out buyers, in news first reported by the Financial Times.
The AA’s owners are understood to value it at £5bn or more. They are also considering floating it on the London stock market – a decade on from its flotation by previous private equity owners. The plans are still in the early stages.
Andrew Yang’s revived pitch suits the automation debate, but UBI can’t fix inequalities concentrated tech wealth drives
Universal basic income (UBI) is back, like a space zombie in a sci-fi movie, resurrected from policy oblivion, hungry for policymakers’ attention: brains!
Andrew Yang, whose “Yang Gang” enthusiasm briefly shook up the Democratic presidential nomination in 2020 promoting a “Freedom Dividend” to save workers from automation – $1,000 a month for every American adult – is again the main carrier of the bug: offering UBI to save the nation when robots eat all our jobs.
The US company behind the Roomba robot vacuum cleaner has filed for bankruptcy protection and agreed to be taken over by one of its Chinese suppliers.
iRobot, which is best known for debuting the Roomba vacuum cleaner in the early 2000s, will be taken over by a subsidiary of its main supplier, Picea Robotics.
House prices in the UK could rise by as much as 4% next year but getting on the property ladder may become slightly less difficult, according to forecasts from the lender Nationwide.
Robert Gardner, the chief economist at the building society, said prices were likely to increase by 2-4%. “We expect housing market activity to strengthen a little further, as affordability improves gradually via income growth outpacing house price growth and a further modest decline in interest rates,” he said.
Depending who you ask, AI-powered coding is either giving software developers an unprecedented productivity boost or churning out masses of poorly designed code that saps their attention and sets software projects up for serious long term-maintenance problems.
The problem is right now, it’s not easy to know which is true.
As tech giants pour billions into large language models (LLMs), coding has been touted as the technology’s killer app. Both Microsoft CEO Satya Nadella and Google CEO Sundar Pichai have claimed that around a quarter of their companies’ code is now AI-generated. And in March, Anthropic’s CEO, Dario Amodei, predicted that within six months 90% of all code would be written by AI. It’s an appealing and obvious use case. Code is a form of language, we need lots of it, and it’s expensive to produce manually. It’s also easy to tell if it works—run a program and it’s immediately evident whether it’s functional.
This story is part of MIT Technology Review’s Hype Correction package, a series that resets expectations about what AI is, what it makes possible, and where we go next.
Executives enamored with the potential to break through human bottlenecks are pushing engineers to lean into an AI-powered future. But after speaking to more than 30 developers, technology executives, analysts, and researchers, MIT Technology Review found that the picture is not as straightforward as it might seem.
For some developers on the front lines, initial enthusiasm is waning as they bump up against the technology’s limitations. And as a growing body of research suggests that the claimed productivity gains may be illusory, some are questioning whether the emperor is wearing any clothes.
The pace of progress is complicating the picture, though. A steady drumbeat of new model releases mean these tools’ capabilities and quirks are constantly evolving. And their utility often depends on the tasks they are applied to and the organizational structures built around them. All of this leaves developers navigating confusing gaps between expectation and reality.
Is it the best of times or the worst of times (to channel Dickens) for AI coding? Maybe both.
A fast-moving field
It’s hard to avoid AI coding tools these days. There are a dizzying array of products available, both from model developers like Anthropic, OpenAI, and Google and from companies like Cursor and Windsurf, which wrap these models in polished code-editing software. And according to Stack Overflow’s 2025 Developer Survey, they’re being adopted rapidly, with 65% of developers now using them at least weekly.
AI coding tools first emerged around 2016 but were supercharged with the arrival of LLMs. Early versions functioned as little more than autocomplete for programmers, suggesting what to type next. Today they can analyze entire code bases, edit across files, fix bugs, and even generate documentation explaining how the code works. All this is guided through natural-language prompts via a chat interface.
“Agents”—autonomous LLM-powered coding tools that can take a high-level plan and build entire programs independently—represent the latest frontier in AI coding. This leap was enabled by the latest reasoning models, which can tackle complex problems step by step and, crucially, access external tools to complete tasks. “This is how the model is able to code, as opposed to just talk about coding,” says Boris Cherny, head of Claude Code, Anthropic’s coding agent.
These agents have made impressive progress on software engineering benchmarks—standardized tests that measure model performance. When OpenAI introduced the SWE-bench Verified benchmark in August 2024, offering a way to evaluate agents’ success at fixing real bugs in open-source repositories, the top model solved just 33% of issues. A year later, leading models consistently score above 70%.
In February, Andrej Karpathy, a founding member of OpenAI and former director of AI at Tesla, coined the term “vibe coding”—meaning an approach where people describe software in natural language and let AI write, refine, and debug the code. Social media abounds with developers who have bought into this vision, claiming massive productivity boosts.
But while some developers and companies report such productivity gains, the hard evidence is more mixed. Early studies from GitHub, Google, and Microsoft—all vendors of AI tools—found developers completing tasks 20% to 55% faster. But a September report from the consultancy Bain & Company described real-world savings as “unremarkable.”
Data from the developer analytics firm GitClear shows that most engineers are producing roughly 10% more durable code—code that isn’t deleted or rewritten within weeks—since 2022, likely thanks to AI. But that gain has come with sharp declines in several measures of code quality. Stack Overflow’s survey also found trust and positive sentiment toward AI tools falling significantly for the first time. And most provocatively, a July study by the nonprofit research organization Model Evaluation & Threat Research (METR) showed that while experienced developers believed AI made them 20% faster, objective tests showed they were actually 19% slower.
Growing disillusionment
For Mike Judge, principal developer at the software consultancy Substantial, the METR study struck a nerve. He was an enthusiastic early adopter of AI tools, but over time he grew frustrated with their limitations and the modest boost they brought to his productivity. “I was complaining to people because I was like, ‘It’s helping me but I can’t figure out how to make it really help me a lot,’” he says. “I kept feeling like the AI was really dumb, but maybe I could trick it into being smart if I found the right magic incantation.”
When asked by a friend, Judge had estimated the tools were providing a roughly 25% speedup. So when he saw similar estimates attributed to developers in the METR study he decided to test his own. For six weeks, he guessed how long a task would take, flipped a coin to decide whether to use AI or code manually, and timed himself. To his surprise, AI slowed him down by an median of 21%—mirroring the METR results.
This got Judge crunching the numbers. If these tools were really speeding developers up, he reasoned, you should see a massive boom in new apps, website registrations, video games, and projects on GitHub. He spent hours and several hundred dollars analyzing all the publicly available data and found flat lines everywhere.
“Shouldn’t this be going up and to the right?” says Judge. “Where’s the hockey stick on any of these graphs? I thought everybody was so extraordinarily productive.” The obvious conclusion, he says, is that AI tools provide little productivity boost for most developers.
Developers interviewed by MIT Technology Review generally agree on where AI tools excel: producing “boilerplate code” (reusable chunks of code repeated in multiple places with little modification), writing tests, fixing bugs, and explaining unfamiliar code to new developers. Several noted that AI helps overcome the “blank page problem” by offering an imperfect first stab to get a developer’s creative juices flowing. It can also let nontechnical colleagues quickly prototype software features, easing the load on already overworked engineers.
These tasks can be tedious, and developers are typically glad to hand them off. But they represent only a small part of an experienced engineer’s workload. For the more complex problems where engineers really earn their bread, many developers told MIT Technology Review, the tools face significant hurdles.
Perhaps the biggest problem is that LLMs can hold only a limited amount of information in their “context window”—essentially their working memory. This means they struggle to parse large code bases and are prone to forgetting what they’re doing on longer tasks. “It gets really nearsighted—it’ll only look at the thing that’s right in front of it,” says Judge. “And if you tell it to do a dozen things, it’ll do 11 of them and just forget that last one.”
DEREK BRAHNEY
LLMs’ myopia can lead to headaches for human coders. While an LLM-generated response to a problem may work in isolation, software is made up of hundreds of interconnected modules. If these aren’t built with consideration for other parts of the software, it can quickly lead to a tangled, inconsistent code base that’s hard for humans to parse and, more important, to maintain.
Developers have traditionally addressed this by following conventions—loosely defined coding guidelines that differ widely between projects and teams. “AI has this overwhelming tendency to not understand what the existing conventions are within a repository,” says Bill Harding, the CEO of GitClear. “And so it is very likely to come up with its own slightly different version of how to solve a problem.”
The models also just get things wrong. Like all LLMs, coding models are prone to “hallucinating”—it’s an issue built into how they work. But because the code they output looks so polished, errors can be difficult to detect, says James Liu, director of software engineering at the advertising technology company Mediaocean. Put all these flaws together, and using these tools can feel a lot like pulling a lever on a one-armed bandit. “Some projects you get a 20x improvement in terms of speed or efficiency,” says Liu. “On other things, it just falls flat on its face, and you spend all this time trying to coax it into granting you the wish that you wanted and it’s just not going to.”
Judge suspects this is why engineers often overestimate productivity gains. “You remember the jackpots. You don’t remember sitting there plugging tokens into the slot machine for two hours,” he says.
And it can be particularly pernicious if the developer is unfamiliar with the task. Judge remembers getting AI to help set up a Microsoft cloud service called an Azure Functions, which he’d never used before. He thought it would take about two hours, but nine hours later he threw in the towel. “It kept leading me down these rabbit holes and I didn’t know enough about the topic to be able to tell it ‘Hey, this is nonsensical,’” he says.
The debt begins to mount up
Developers constantly make trade-offs between speed of development and the maintainability of their code—creating what’s known as “technical debt,” says Geoffrey G. Parker, professor of engineering innovation at Dartmouth College. Each shortcut adds complexity and makes the code base harder to manage, accruing “interest” that must eventually be repaid by restructuring the code. As this debt piles up, adding new features and maintaining the software becomes slower and more difficult.
Accumulating technical debt is inevitable in most projects, but AI tools make it much easier for time-pressured engineers to cut corners, says GitClear’s Harding. And GitClear’s data suggests this is happening at scale. Since 2020, the company has seen a significant rise in the amount of copy-pasted code—an indicator that developers are reusing more code snippets, most likely based on AI suggestions—and an even bigger decline in the amount of code moved from one place to another, which happens when developers clean up their code base.
And as models improve, the code they produce is becoming increasingly verbose and complex, says Tariq Shaukat, CEO of Sonar, which makes tools for checking code quality. This is driving down the number of obvious bugs and security vulnerabilities, he says, but at the cost of increasing the number of “code smells”—harder-to-pinpoint flaws that lead to maintenance problems and technical debt.
Recent research by Sonar found that these make up more than 90% of the issues found in code generated by leading AI models. “Issues that are easy to spot are disappearing, and what’s left are much more complex issues that take a while to find,” says Shaukat. “That’s what worries us about this space at the moment. You’re almost being lulled into a false sense of security.”
If AI tools make it increasingly difficult to maintain code, that could have significant security implications, says Jessica Ji, a security researcher at Georgetown University. “The harder it is to update things and fix things, the more likely a code base or any given chunk of code is to become insecure over time,” says Ji.
There are also more specific security concerns, she says. Researchers have discovered a worrying class of hallucinations where models reference nonexistent software packages in their code. Attackers can exploit this by creating packages with those names that harbor vulnerabilities, which the model or developer may then unwittingly incorporate into software.
LLMs are also vulnerable to “data-poisoning attacks,” where hackers seed the publicly available data sets models train on with data that alters the model’s behavior in undesirable ways, such as generating insecure code when triggered by specific phrases. In October, research by Anthropic found that as few as 250 malicious documents can introduce this kind of back door into an LLM regardless of its size.
The converted
Despite these issues, though, there’s probably no turning back. “Odds are that writing every line of code on a keyboard by hand—those days are quickly slipping behind us,” says Kyle Daigle, chief operating officer at the Microsoft-owned code-hosting platform GitHub, which produces a popular AI-powered tool called Copilot (not to be confused with the Microsoft product of the same name).
The Stack Overflow report found that despite growing distrust in the technology, usage has increased rapidly and consistently over the past three years. Erin Yepis, a senior analyst at Stack Overflow, says this suggests that engineers are taking advantage of the tools with a clear-eyed view of the risks. The report also found that frequent users tend to be more enthusiastic and more than half of developers are not using the latest coding agents, perhaps explaining why many remain underwhelmed by the technology.
Those latest tools can be a revelation. Trevor Dilley, CTO at the software development agency Twenty20 Ideas, says he had found some value in AI editors’ autocomplete functions, but when he tried anything more complex it would “fail catastrophically.” Then in March, while on vacation with his family, he set the newly released Claude Code to work on one of his hobby projects. It completed a four-hour task in two minutes, and the code was better than what he would have written.
“I was like, Whoa,” he says. “That, for me, was the moment, really. There’s no going back from here.” Dilley has since cofounded a startup called DevSwarm, which is creating software that can marshal multiple agents to work in parallel on a piece of software.
The challenge, says Armin Ronacher, a prominent open-source developer, is that the learning curve for these tools is shallow but long. Until March he’d remained unimpressed by AI tools, but after leaving his job at the software company Sentry in April to launch a startup, he started experimenting with agents. “I basically spent a lot of months doing nothing but this,” he says. “Now, 90% of the code that I write is AI-generated.”
Getting to that point involved extensive trial and error, to figure out which problems tend to trip the tools up and which they can handle efficiently. Today’s models can tackle most coding tasks with the right guardrails, says Ronacher, but these can be very task and project specific.
To get the most out of these tools, developers must surrender control over individual lines of code and focus on the overall software architecture, says Nico Westerdale, chief technology officer at the veterinary staffing company IndeVets. He recently built a data science platform 100,000 lines of code long almost exclusively by prompting models rather than writing the code himself.
Westerdale’s process starts with an extended conversation with the modelagent to develop a detailed plan for what to build and how. He then guides it through each step. It rarely gets things right on the first try and needs constant wrangling, but if you force it to stick to well-defined design patterns, the models can produce high-quality, easily maintainable code, says Westerdale. He reviews every line, and the code is as good as anything he’s ever produced, he says: “I’ve just found it absolutely revolutionary,. It’s also frustrating, difficult, a different way of thinking, and we’re only just getting used to it.”
But while individual developers are learning how to use these tools effectively, getting consistent results across a large engineering team is significantly harder. AI tools amplify both the good and bad aspects of your engineering culture, says Ryan J. Salva, senior director of product management at Google. With strong processes, clear coding patterns, and well-defined best practices, these tools can shine.
DEREK BRAHNEY
But if your development process is disorganized, they’ll only magnify the problems. It’s also essential to codify that institutional knowledge so the models can draw on it effectively. “A lot of work needs to be done to help build up context and get the tribal knowledge out of our heads,” he says.
The cryptocurrency exchange Coinbase has been vocal about its adoption of AI tools. CEO Brian Armstrong made headlines in August when he revealed that the company had fired staff unwilling to adopt AI tools. But Coinbase’s head of platform, Rob Witoff, tells MIT Technology Review that while they’ve seen massive productivity gains in some areas, the impact has been patchy. For simpler tasks like restructuring the code base and writing tests, AI-powered workflows have achieved speedups of up to 90%. But gains are more modest for other tasks, and the disruption caused by overhauling existing processes often counteracts the increased coding speed, says Witoff.
One factor is that AI tools let junior developers produce far more code,. As in almost all engineering teams, this code has to be reviewed by others, normally more senior developers, to catch bugs and ensure it meets quality standards. But the sheer volume of code now being churned out i whichs quickly saturatinges the ability of midlevel staff to review changes. “This is the cycle we’re going through almost every month, where we automate a new thing lower down in the stack, which brings more pressure higher up in the stack,” he says. “Then we’re looking at applying automation to that higher-up piece.”
Developers also spend only 20% to 40% of their time coding, says Jue Wang, a partner at Bain, so even a significant speedup there often translates to more modest overall gains. Developers spend the rest of their time analyzing software problems and dealing with customer feedback, product strategy, and administrative tasks. To get significant efficiency boosts, companies may need to apply generative AI to all these other processes too, says Jue, and that is still in the works.
Rapid evolution
Programming with agents is a dramatic departure from previous working practices, though, so it’s not surprising companies are facing some teething issues. These are also very new products that are changing by the day. “Every couple months the model improves, and there’s a big step change in the model’s coding capabilities and you have to get recalibrated,” says Anthropic’s Cherny.
For example, in June Anthropic introduced a built-in planning mode to Claude; it has since been replicated by other providers. In October, the company also enabled Claude to ask users questions when it needs more context or faces multiple possible solutions, which Cherny says helps it avoid the tendency to simply assume which path is the best way forward.
Most significant, Anthropic has added features that make Claude better at managing its own context. When it nears the limits of its working memory, it summarizes key details and uses them to start a new context window, effectively giving it an “infinite” one, says Cherny. Claude can also invoke sub-agents to work on smaller tasks, so it no longer has to hold all aspects of the project in its own head. The company claims that its latest model, Claude 4.5 Sonnet, can now code autonomously for more than 30 hours without major performance degradation.
Novel approaches to software development could also sidestep coding agents’ other flaws. MIT professor Max Tegmark has introduced something he calls “vericoding,” which could allow agents to produce entirely bug-free code from a natural-language description. It builds on an approach known as “formal verification,” where developers create a mathematical model of their software that can prove incontrovertibly that it functions correctly. This approach is used in high-stakes areas like flight-control systems and cryptographic libraries, but it remains costly and time-consuming, limiting its broader use.
Rapid improvements in LLMs’ mathematical capabilities have opened up the tantalizing possibility of models that produce not only software but the mathematical proof that it’s bug free, says Tegmark. “You just give the specification, and the AI comes back with provably correct code,” he says. “You don’t have to touch the code. You don’t even have to ever look at the code.”
When tested on about 2,000 vericoding problems in Dafny—a language designed for formal verification—the best LLMs solved over 60%, according to non-peer-reviewed research by Tegmark’s group. This was achieved with off-the-shelf LLMs, and Tegmark expects that training specifically for vericoding could improve scores rapidly.
And counterintuitively, the speed at which AI generates code could actually ease maintainability concerns. Alex Worden, principal engineer at the business software giant Intuit, notes that maintenance is often difficult because engineers reuse components across projects, creating a tangle of dependencies where one change triggers cascading effects across the code base. Reusing code used to save developers time, but in a world where AI can produce hundreds of lines of code in seconds, that imperative has gone, says Worden.
Instead, he advocates for “disposable code,” where each component is generated independently by AI without regard for whether it follows design patterns or conventions. They are then connected via APIs—sets of rules that let components request information or services from each other. Each component’s inner workings are not dependent on other parts of the code base, making it possible to rip them out and replace them without wider impact, says Worden.
“The industry is still concerned about humans maintaining AI-generated code,” he says. “I question how long humans will look at or care about code.”
A narrowing talent pipeline
For the foreseeable future, though, humans will still need to understand and maintain the code that underpins their projects. And one of the most pernicious side effects of AI tools may be a shrinking pool of people capable of doing so.
Early evidence suggests that fears around the job-destroying effects of AI may be justified. A recent Stanford University study found that employment among software developers aged 22 to 25 fell nearly 20% between 2022 and 2025, coinciding with the rise of AI-powered coding tools.
Experienced developers could face difficulties too. Luciano Nooijen, an engineer at the video-game infrastructure developer Companion Group, used AI tools heavily in his day job, where they were provided for free. But when he began a side project without access to those tools, he found himself struggling with tasks that previously came naturally. “I was feeling so stupid because things that used to be instinct became manual, sometimes even cumbersome,” says Nooijen.
Just as athletes still perform basic drills, he thinks the only way to maintain an instinct for coding is to regularly practice the grunt work. That’s why he’s largely abandoned AI tools, though he admits that deeper motivations are also at play.
Part of the reason Nooijen and other developers MIT Technology Review spoke to are pushing back against AI tools is a sense that they are hollowing out the parts of their jobs that they love. “I got into software engineering because I like working with computers. I like making machines do things that I want,” Nooijen says. “It’s just not fun sitting there with my work being done for me.”
Britain’s financial watchdog is setting out plans today to reform the UK mortgage market to help first-time buyers and the self-employed onto the housing ladder.
The Financial Conduct Authority (FCA) is proposing to improve four areas of the mortgage market – including rule changes to allow lenders to offer ‘more flexible’ mortages.
First-time buyers & underserved consumers: Simplifying mortgage rules to allow more flexible products that reflect different working patterns and income levels at different stages of life.
Later-life lending: Reviewing retirement interest-only requirements to make them more accessible. Exploring ways to improve advice to help people confidently plan for later life. Conducting a focused market study to ensure the lifetime mortgage market can meet the changing needs of future customers.
Innovation & disclosure: Encouraging the use of data and technology, such as AI, to help brokers give better and faster advice while keeping a human touch. Looking at ways to make advertising and disclosure rules simpler, so consumers can understand information online more easily.
Protecting vulnerable consumers: Working with partners to support people affected by financial abuse and help those using a mortgage to manage or consolidate debt.
Resolution Foundation report comes in week when data is expected to show October unemployment rise
Young people are bearing the brunt of Britain’s jobs downturn, according to a report, before official figures this week that are expected to show the UK unemployment rate rising to 5.1%.
The Resolution Foundation thinktank said a “jobs deficit” was pushing a growing number of graduates and non-graduates into unemployment as employers reduced hiring.
Japanese green tea named stain of the year as survey finds Aperol spritz and bubble tea are also leaving their mark
It used to be curry sauce, egg yolk and red wine that ruined Britain’s clothes but in a sign of the times laundry detergents are being reformulated to tackle stains left by matcha lattes, Aperol spritz and bubble tea.
In a month when year-end gongs are dished out, from BBC Sports Personality to Pantone’s Colour of 2026 (a white called “cloud dancer”), matcha has received the dubious accolade “stain of the year”.
People able to buy homes previously beyond budget, aided by rising wages and looser affordability tests
First-time buyers are taking out larger mortgages than ever before as rising wages and looser affordability tests allow them to buy properties that were previously beyond their budget.
The average first-time buyer borrowed £210,800 in the year to September, a record high, according to analysis by Savills, the property agent.
Details of financing structure to be reviewed by culture secretary and regulators before deal can proceed
The owner of the Daily Mail has secured funding for a £500m takeover of the Telegraph, in a crucial development that paves the way for the group to announce the terms of its acquisition on Monday.
Lord Rothermere’s Daily Mail and General Trust (DMGT) has agreed to pay the sum in two instalments, according to weekend reports. An initial payment of £400m will be funded by an increase in the group’s debt with its longstanding lender NatWest and existing company cash.
From choosing frozen turkey to shopping supermarket specials, the traditional meal need not break the household budget
Figures show that the total cost of the all-important Christmas dinner is up 5% on a year ago, with the price of important elements such as pigs in blankets and stuffing up by 7%.
With the cost of living still biting, however, a supermarket price war is taking some of the sting out of high food costs – with Aldi and Lidl selling the ingredients for a main Christmas meal for eight for less than £12.
The City expects a cut in interest rates on Thursday but the economic prospects for 2026 complicate the picture
In the economic gloom of Labour’s first year in power, Rachel Reeves has had a reliable shred of comfort to cling to: five times since the general election, the Bank of England has cut interest rates.
This week, in all likelihood, the chancellor will get a sixth to shout about, as Threadneedle Street prepares to reduce borrowing costs in an early Christmas present that will be seized upon by the Treasury.
With public service broadcasters starting to look like ‘endangered species’, many want Channel 4 and BBC to work more closely
The prospect of Comcast taking over ITV has prompted concerns about the impact on British public service broadcasting, a fact that Channel 4’s new chief executive, moving from a senior post at Sky, will be all too well aware.
Sky’s advertising chief, Priya Dogra, will now be expected to lead the charge to block her former employer’s takeover plan to protect Channel 4.
Training and qualifications body, acquired by private Greek firm in October, to become ‘leaner organisation’
The training and qualifications body City & Guilds is shrinking its UK workforce as part of a £22m cost-cutting drive after it was acquired by a private Greek business in October.
Founded in 1878 by the City of London and a group of 16 livery companies, the original institute developed a national system of technical education, offering qualifications and apprenticeships in fields ranging from manufacturing and mechanical engineering to hairdressing and horticulture.
Fossil fuel execs Robert Pender and Michael Sabel deny wrongdoing after report on potential conflict of interest
Two more senior Democrats have called for an investigation into a share-buying spree by two fossil fuel billionaires with close ties to the Trump administration, after a Guardian investigation raised questions about potential wrongdoing.
Robert Pender and Michael Sabel, the founders and co-chairs of Venture Global, a liquefied natural gas (LNG) company headquartered in Virginia, bought more than a million shares worth almost $12m each in March. The trades took place just days after a meeting with senior White House officials, who then issued a key regulatory permit that helped expand the company’s business in Europe.
Drivers also told to take caution after complaints rise about long-stay meet-and-greet services
Terence Baxter* had booked a meet-and-greet service to park his Volkswagen at Heathrow airport while he and his wife went on holiday. The couple handed over the keys at the drop-off site and were driven to the terminal – and that was the last they saw of their car. On their return they were informed by the company it had been stolen.
Their case comes as airports and police forces are warning travellers to be wary of “unofficial” operators advertising cheap long-stay parking after a rise in complaints.
Several car-sharing companies are considering launching or expanding in London, with the imminent closure of Zipcar’s UK operation leaving a large gap in the market in one of Europe’s biggest cities.
Free2Move, owned by the carmaker Stellantis, said it was “closely monitoring the London market”, and “actively assessing” options for its services. It already operates fleets in cities including Berlin, Paris, Rome and Washington DC.
California jury finds company knew its talc-based products were dangerous but failed to warn consumers
A California jury on Friday awarded $40m to two women who said Johnson & Johnson’s baby powder was to blame for their ovarian cancer.
The jury in Los Angeles superior court awarded $18m to Monica Kent and $22m to Deborah Schultz and her husband after finding that Johnson & Johnson knew for years its talc-based products were dangerous but failed to warn consumers.
id Software employees voted to form a wall-to-wall union with the CWA, covering all roles at the Doom studio. "The vote wasn't unanimous, though a majority did vote in favor of the union," notes Engadget. From the report: The union will work in conjunction with the Communications Workers of America (CWA), which is the same organization involved with parent company ZeniMax's recent unionization efforts. Microsoft, who owns ZeniMax, has already recognized this new effort, according to a statement by the CWA. It agreed to a labor neutrality agreement with the CWA and ZeniMax workers last year, paving the way for this sort of thing.
From the onset, this union will look to protect remote work for id Software employees. "Remote work isn't a perk. It's a necessity for our health, our families, and our access needs. RTO policies should not be handed down from executives with no consideration for accessibility or our well-being," said id Software Lead Services Programmer Chris Hays. He also said he looks forward to getting worker protections regarding the "responsible use of AI."
Login
