The CEO of artificial ielligence startup Ahropic says companies in the AI industry must be transpare about the risks of their products, or they risk making the same mistakes that befell cigarette, tobacco and opioid companies.
Dario Amodei, CEO of Ahropic, told CBS News that he believes AI will become smarter than all or most humans in all or most aspects. He asked other managers of artificial ielligence companies to be frank and honest in their actions.
Amudi explained that the lack of transparency around the effects of artificial ielligence can repeat the same mistakes made by cigarette and opioid companies and cause the inability to explain the possible harms of these products and sound the alarm in this field.
“You could end up in a world like cigarette or opioid companies, where they knew there were risks, but they didn’t talk about them, and they certainly didn’t preve them,” said Eropic’s CEO.
Ahropic CEO warns about AI capabilities
Dario Amudi, who has previously warned about the impact of artificial ielligence on jobs such as accouing, law and banking, explained: “Without hard ierveion, it is conceivable that there will not be a big impact on the world of business. “I’m concerned that the impact will be broader and faster than what we’ve seen with past technologies.”
The CEO of Ahropic said about the dangers of artificial ielligence in terms of extorting users and conducting cyber attacks: “One of the strengths of models is that they can act arbitrarily. “But the more we empower these systems, the more we can worry about whether they’re doing exactly what we wa them to do.”
If a model can help make a biological weapon, it means the level of its capabilities has reached a poi where it can help make a vaccine or develop drugs more quickly, Amody says. Therefore, it seems that the role of model makers is esseial in guiding these technologies.
“You wa a model that will build you a business and make billions in revenue,” he explained. But you don’t wa to wake up one day and find yourself out of your company. “So our basic approach is to evaluate these autonomous abilities and do as many weird experimes as possible and see what happens.”
