A former security employee at OpenAI, the company behind ChatGPT, has criticized the approach of AI companies to managing the risks that arise when developing increasingly sophisticated systems. David Robinson believes that accelerating the technological race increases the probability of failures and requires a more cautious stance.
In an article published this Saturday (3) in the American magazine The Atlantic, Robinson stated that AI development companies, including OpenAI, are 'far from cautious enough.'
According to the former employee, companies should pay more attention to safety specialists and research before beginning the development of more advanced systems. Robinson emphasized: 'The time for trial and error is over.'
He argues that advanced AI systems must have safety measures similar to those used in sectors such as nuclear energy and aviation. Robinson also criticized OpenAI's practice, which it calls 'iterative deployment,' involving launching systems and subsequently strengthening safety measures as problems are discovered.
These remarks are part of the ongoing AI discussion regarding the speed at which companies are creating increasingly complex systems. OpenAI and its competitor Anthropic are facing scrutiny following incidents where safety systems failed or experimental systems exhibited unexpected behavior.
More information:
Robinson's criticism is not an isolated incident among specialists who have worked at AI companies. Last month, a former researcher who left OpenAI to work at Anthropic also left the industry, accusing both American companies of 'playing with our lives.'
Jacob Coxon, a 27-year-old Briton, worked for three years on pre-training—the stage where AI models absorb vast amounts of information. He first worked at OpenAI and then at Anthropic this year.
Coxon wrote on X: 'Neither company is acting responsibly. They are rushing straight toward superintelligence capable of self-improvement, putting our lives at risk.'