AI was supposed to speed up coders, new study says it did the opposite
The findings challenge the widespread notion that AI tools automatically make human coders more efficient, a belief that has attracted billions of dollars in investment and sparked predictions that AI could soon replace many junior engineering roles.Past studies have shown strong productivity gains with AI. One found that AI helped developers complete 56 per cent more code, while another claimed a 26 per cent boost in task volume. But the METR study suggests that those gains don't apply to all situations, especially where developers already have deep familiarity with the code.Instead of streamlining work, the AI often made suggestions that were only 'directionally correct,' said Becker. 'When we watched the videos, we found that the AIs made some suggestions about their work, and the suggestions were often directionally correct, but not exactly what's needed.'As a result, developers spent additional time reviewing and correcting AI-generated code, which ultimately slowed them down. However, the researchers do not believe this slowdown would apply to all coding scenarios, such as those involving junior developers or unfamiliar codebases.Despite the results, both the study's authors and most participants continue to use Cursor. Becker suggested that while the tool may not speed up work, it can still make development feel easier and more enjoyable.'Developers have goals other than completing the task as soon as possible,' he said. 'So they're going with this less effortful route.'The authors also emphasised that their findings should not be over-generalised. The slowdown only reflects a snapshot of AI's capabilities as of early 2025, and further improvements in prompting, training, and tool design could lead to different outcomes in future.As AI systems continue to evolve, METR plans to repeat such studies to better understand how AI might accelerate, or hinder, human productivity in real-world development settings.- Ends
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Time of India
40 minutes ago
- Time of India
TCS CEO K. Krithivasan's net worth: Amid 12,000 job cuts, top IT executive earns Rs 265200000 remuneration annually
In another major move within the IT sector, India's largest services provider company, Tata Consultancy Services (TCS), is aiming to reduce its workforce by 2 per cent or approximately 12,000 employees, in its 2026 financial year, as per reports that came in on Sunday, July 27, 2025. Amid the TCS layoffs, let's explore the net worth of CEO K. Krithivasan and understand his financial standing. According to Mint, the company is retraining and redeploying staff in a bid to become more agile and more firm to be ready for the future amid the fears of disruptions in technology, especially due to the emergence of Artificial Intelligence (AI). Moreover, the company's target is primarily focused on the middle and senior management, as per Reuters. However, this decision will also impact the employees from across the world and the domains where TCS operates. TCS to lay off 12,000 employees.— Tata Consultancy Services will lay off 2% of its workforce over the next year, Moneycontrol reported.— The move will impact around 12,000 of the company's employees, primarily at middle and senior levels, TCS CEO K Krithivasan told… Who is K. Krithivasan? K. Krithivasan was appointed as the CEO of Tata Consultancy Services (TCS) after MD and CEO Rajesh Gopinathan stepped down from his role in 2023. Krithivasan is currently the president and global head of the banking, financial services, and insurance business unit at TCS. According to an excerpt shared on the company's website, "He is responsible for planning and executing growth strategies, improving financial performance, and enhancing customer mindshare and market positioning. He has helped key clients with digital transformation, change management cycle acceleration, achieving value beyond cost optimisation, and establishing IT program governance." K. Krithivasan's net worth and more The recently revealed annual report of IT services TCS provided a valuable insight into the C-suite executive salaries in India. According to The Week, the annual report revealed that K. Krithivasan takes home a staggering Rs 26.52 crore in FY2025, which is ₹2.21 crore a month! The report further mentioned, in the fiscal year 2025, the CEO's salary increased by 4.6 per cent, and this stands in stark contrast to the median salary at TCS, as Krithivasan earned nearly 330 times more than others, which highlighted the significant salary gap within the IT services industry. His total compensation of Rs 26.52 crore consisted of a Rs 1.39 crore salary, Rs 2.13 crore in benefits (including perks and allowances), and a commission of Rs 23 crore, as detailed in the annual report. FAQs Q. Why is TCS cutting 2% of its employees in the fiscal year 2026? Given the possible disruptions from cutting-edge technologies like artificial intelligence, TCS wants to become more adaptable and ready for any obstacles that may arise in the future. Q. How much does K. Krithivasan get paid in comparison to other CEOs in the IT industry in India? According to Financial Express, K. Krithivasan's total compensation of ₹26.52 crore for the fiscal year 2024–25 was much less than that of his colleagues. The CEOs of HCLTech, Infosys, and Wipro, for example, received ₹84.16 crore, ₹80.6 crore, and almost ₹53 crore, respectively. For the latest and more interesting financial news, keep reading Indiatimes Worth. Click here.
Mint
4 hours ago
- Mint
TCS layoffs: IT major to cut over 12,000 jobs, middle and senior management to get affected. Here's what we know
TCS layoffs: India's largest software services provider Tata Consultancy Services (TCS) will reduce its workforce by 2% or roughly over 12,000 employees in its 2026 financial year. The company is retraining and redeploying staff in a bid to become more agile and future ready amid rapid disruptions in technology, particularly around Artificial Intelligence (AI). The company's decision is primarily targeted at the middle and senior management, reported Reuters. This decision will impact employees from different countries and domains where TCS operates. TCS has an employee headcount of 6,13,000 for the latest quarter ended June. TCS assured that this transition is being planned with 'due care' to ensure no disruption in delivering service to the clients. In the April-June quarter, TCS's operating margins narrowed 20 basis points to 24.5%, amid the company's plans to boost them to 26-28% levels. Just two weeks ago, TCS also said that their 'priority' focus is delivering wage hikes for its over 6 lakh workforce. 'My priority is getting back to the wage hike,' said Chief Financial Officer (CFO) Samir Seksaria without specifying a deadline for the salary hike for its employees. TCS employee attrition rate climbed 13.8% on a last twelve-month (LTM) basis as of the end of the April to June quarter of FY2025-26. This marked a marginal rise on a quarter-on-quarter (QoQ) basis, compared to its 13.3% levels in the January-March quarter of FY 2024-25. The company's CFO said that with the attrition rate hitting a 'concerning level' as of the April-June quarter, the company will now focus on retaining top-level talent, which is difficult to build through fresh hiring. He also hinted that as the company has capacity now, it may go slow on lateral hirings and restart once the demand surges, according to the agency report.
Economic Times
4 hours ago
- Economic Times
You can still outpace AI: Wharton professor reveals a ‘skill bundling' strategy to safeguard your future from automation
iStock Artificial Intelligence is rapidly changing workplaces. Wharton professor Ethan Mollick suggests professionals focus on roles requiring diverse human skills. These include emotional intelligence and creativity. (Image: iStock) As artificial intelligence reshapes the modern workplace with stunning speed, one Wharton professor has a sobering message for today's professionals: the safest jobs of tomorrow aren't necessarily the most technical—they're the most complex. Ethan Mollick, associate professor at the Wharton School and author of Co-Intelligence: Living and Working with AI, says job security in the AI era will increasingly depend on choosing roles that bundle multiple human skills together. That means emotional intelligence, judgment, creativity, and domain expertise—all woven into one. 'AI may outperform you in one or two things,' Mollick tells CNBC Make It, 'but if your job requires five or six of them, it's a lot harder to replace.' It's the kind of insight that redefines how we think about employability in an increasingly automated world. And with AI usage surging—40% of U.S. workers now use it at least a few times a year, per a Gallup poll—these career choices have never mattered more. Mollick doesn't sugarcoat the AI wave ahead. Tech labs aren't just chasing progress—they're chasing a paradigm shift. 'Labs are aiming for machines smarter than humans within the next three years,' Mollick warns. 'They're betting on mass unemployment. Whether they succeed or not is still unclear, but we have to take it as a real possibility.' Even Nvidia CEO Jensen Huang, whose company powers some of the most advanced AI systems, echoes that sentiment—albeit from a different vantage point. In a recent All-In podcast, Huang predicted AI will create more millionaires in five years than the internet did in 20, while also cautioning: 'Anybody who is not using AI will lose their job to someone who is.' What's the solution? According to Mollick, job seekers must rethink their strategy. 'Don't go for roles that do one thing,' he says. 'Pick a job like being a doctor—where you're expected to be good at empathy, diagnosis, hand skills, and research. If AI helps with some of it, you still have the rest.' This idea of "bundled roles"—where a single job draws on varied skills and responsibilities—could be the firewall against replacement. These complex, human-centered positions are harder for AI to replicate wholesale and leave more room for humans to collaborate with AI, not compete against it. AI's evolution could make entry-level roles scarce—or at least, radically different. 'Companies will need to rethink entry-level hiring,' Mollick notes. 'Not just for productivity, but for training future leaders.' Without the chance to learn through repetition—what Mollick calls 'apprenticeship'—younger workers may miss out on foundational skills. The result could be a workforce with knowledge gaps AI can't fill, even as those same gaps are used to justify greater automation. Nvidia's Huang calls AI the 'greatest equalizer of our time' because it gives creative power to anyone who can express an idea. 'Everybody is a programmer now,' he says. But critics caution that this accessibility may also deepen divides between the AI-literate and those left behind. Eric Schmidt, former Google CEO, has a different concern: infrastructure. On the Moonshots podcast, Schmidt warned that AI's growth could be throttled not by chips, but by electricity. The U.S., he says, may need 92 more gigawatts of power to meet AI demands—equivalent to 92 new nuclear plants. As AI spreads into every corner of work, from payroll review (yes, Huang uses machine learning for that too) to high-stakes decision-making, the one thing that's clear is this: the rules are changing faster than most organizations can adapt. 'The tools are evolving fast,' Mollick says, 'but organizations aren't. And we can't ask employees to figure it all out on their own.' He believes the real danger isn't AI itself—but the lack of vision from leadership. Without a clear roadmap, workers are left adrift, trying to 'magic' their way into the future. In the race to stay relevant in the AI era, the best defense isn't to out-code or out-process a machine. It's to out-human it—by doubling down on the kind of nuanced, multi-layered work AI can't yet replicate. And by choosing jobs that ask you to wear many hats, not just one. Or as Mollick puts it: 'Bundled tasks are your best bet for surviving the AI takeover.'
