Microsoft chief science officer Eric Horvitz believes artificial intelligence (AI) will enable humans to thrive in the future, but he appears to be avoiding the bet.
Mr. Horvitz published a post on the official Microsoft blog on May 30th entitled “Thoughts on AI and the future of human prosperity.” This article discusses the future of artificial intelligence and presents a series of essays written by his AI expert who was granted early access to OpenAI. His GPT-4 before public release.
Horvitz was surprised by OpenAI’s GPT-4 capabilities when he was given such access in 2022, according to the blog post.
“I have observed an unexpected brilliance of intelligence beyond what we have seen in previous AI systems. In comparison, I noticed a significant improvement in functionality.
Specifically, Horvitz touts GPT-4’s “ability to interpret my intentions and provide sophisticated answers to numerous prompts,” which he describes as “an emergent phenomenon encountered in physics. It felt like a ‘phase transition’ that evokes the image of
While testing GPT-4, Horvitz noted that “this model and its successors […] It has great potential for change. ”
Microsoft and OpenAI have provided early access to GPT-4 to many experts and solicited essays from them in hopes of identifying the opportunities and challenges that this technology poses to various industries.
Related: EU, US should promote AI code of conduct: EU technology chief
However, on the same day Horvitz published his announcement post, his name also appeared as a signatory to a document issued by the Center for AI Safety.
The document contains only a preface and a one-sentence statement, signed by a list of signatories that represent true portraits of AI experts (Yann LeCun of Meta, Google Brain).
“Reducing the risk of AI-induced extinction should be a global priority alongside other societal-scale risks such as pandemics and nuclear war,” the statement said.
Reception of the document has been somewhat divided on social media, with many experts detailing reasons for signing, while worrying about undefined risks, such as future AI systems with capabilities not possible today. Some experts worry that it distracts them from more important matters. .
In a recent blog post, Horvitz notes the “challenges” involved in developing secure AI and acknowledges “fears” about the technology’s transformative nature. However, there is no mention of the possibility of extinction.
Beyond AI, Microsoft is researching other early technologies. The company was recently approved to participate in the Central Bank Digital Currency Pilot Project of the Central Bank of Brazil, along with Santander, Visa and several other applicants.
