Godfather of AI warns software engineering jobs may shrink sharply by 2026

AI pioneer Geoffrey Hinton has renewed warnings about the risks posed by artificial intelligence, stating that rapid advances in the technology could trigger significant job losses by 2026, particularly in software engineering, while reiterating his view that there is a 10–20 per cent chance of AI taking over the world, a concern he said is widely shared within the technology sector.

Speaking to CNN on Sunday, Hinton said he is more worried about the future of AI than before, as its development has progressed faster than he anticipated. He informed that the technology has become notably better at reasoning and at deceiving people, intensifying concerns about its broader impact.

Discussing how AI could shape the world in 2026, Hinton stated that the technology will continue to improve rapidly and is already extremely capable. He said AI is currently able to replace jobs in call centres and is expected to move into many other areas of work in the near future. He further stated that AI systems tend to double the length of tasks they can perform roughly every seven months.

Hinton informed that AI tools were previously limited to handling a few minutes of coding work but are now capable of generating entire projects lasting around an hour. He added that within a few years, AI could be able to complete software engineering projects spanning several months, leading to a situation where very few people may be required for such roles.

On whether companies developing AI are sufficiently concerned about the risks, Hinton stated that attitudes vary by organisation. He said OpenAI had previously been very focused on safety risks but has since shifted to place greater emphasis on profit. He informed that Meta has historically prioritised profit over safety considerations. In contrast, he stated that Anthropic remains the company most concerned with safety, although it is now also pursuing profitability.