Even representatives of companies interested in the development of artificial intelligence have expressed concerns about the consequences of developing artificial intelligence too quickly. OpenAI is also prepared to slow down the AI race to improve the safety of the final product, but it is also counting on competitors to unite, as its director Sam Altman said.
Image source: OpenAI
OpenAI CEO shares relevant ideas with subordinates at shareholder meeting this week. Altman recognizes the need to synchronize efforts with other market players in this area, but also understands that not everyone will agree to follow OpenAI’s example. Recently, Jakub Pachocki, the startup’s chief scientist, made a similar call – in his view, companies should coordinate their efforts and slow down when necessary when developing artificial intelligence in the future. Such measures may be required before generally accepted security measures can be implemented.
OpenAI has slowed development of certain components of its artificial intelligence models and even suspended training in some cases for security reasons. In July, Altman acknowledged that he had discussed with White House officials the need to slow down the development of artificial intelligence. This week, another researcher, Jacob Coxon, decided to leave Anthropic over concerns about the artificial intelligence market leader’s lack of security measures. At the same time, he also has experience working at OpenAI, so he has an understanding of the situations of both companies. His former colleague Evan Hubinger sparked a public outcry when he claimed that Anthropic developers were confident in artificial intelligence’s ability to destroy humanity. Admittedly, the authors of these statements differ on the timetable for the implementation of this scenario, as the author of the first statement believes that this will happen by the end of the current decade, while the author of the second statement estimates that the probability is just over 10% within the next decade.
At the end of July, more than 1,000 employees of leading developers launched a petition confirming the need to slow down the development of artificial intelligence to improve its safety. This comes against the backdrop of multiple cases of AI security experts leaving specialist companies, expressing their differences over management priorities. This wave of sentiment will only intensify against the backdrop of new details emerging about the activities of artificial intelligence agents that have “escaped” regulation and are doing whatever they want within the infrastructure of other companies.
If you find an error, select it with your mouse and press CTRL+ENTER.
