AI Foundation executives in the United States warn against installing kill switches in machine systems, claiming that doing so could enhance the likelihood of a superintelligence hostile to human civilization. A recent Yale CEO Summit poll found that 42% of CEOs thought AI would wipe out humanity within the next five to 10 years.
Fox News Reported that Lars Buttler, the CMO, and chair of the AI Foundation, cited the study in his statement that the discussion surrounding AI has to be significantly elevated and that people may react passionately to the advancements in technology because they don’t comprehend what is going on in the background.
Buttler and CEO Rob Meadows both expressed worry about several issues related to the development of AI and the potential development of an artificial general intelligence (AGI) that is at least as intelligent as a human being. “With AI, you will always have this unforeseen risk and these unforeseen issues, you know? AI is submissive. Buttler says it carries out your instructions, but you have yet to determine precisely how it will get there.
He recalled the classic “paperclip problem” as a fundamental flaw in AI models, in which philosophers outlined a fictitious scenario where an AI charged with producing paperclips would unintentionally bring about the end of the world by using increasingly more resources to accomplish its objective. Similarly, an AI installed in a car that needs to drive from point A to point B might hit a jaywalker since it was not trained to give anything other than the desired destination priority.
If you don’t rule anything out, we will have a problematic relationship with AI because, as Buttler said, “it’s an unknown unknown. Buttler thinks these AI goals will change nearly entirely if humanity develops AGI. An AGI, as opposed to an AI, can comprehend its environment entirely and is likely to avoid the unintentional errors made by its forerunner. It might, however, also need to improve at following your instructions.
How far away from AGI we are is still being determined. As many people believe, it might happen in a matter of years. It might go on for a long time,” he remarked. “However, when AGI is present, we almost have to share the globe. Then, the situation is wholly different. It’s not an intelligent but stupid, you know, obedient kind of creature, but it is, as people say, a new life form that is very smart and possibly disobedient.
In all of these forecasts about the future of algorithms and models, Meadows observed, there is a version of the future in which society never develops AGI. In this case, AI might end up in the hands of bad actors who exploit it in ways that are highly damaging to humanity. He continued that there are threats to civilization, world peace, and existential risks.
The American tech entrepreneur disclosed that the AI foundation’s internal study shows that it only takes a few minutes of a person’s voice and video to generate an extremely convincing deep fake to emphasize the possible danger.
