Eric Horvitz expressed admiration for the transformative potential of AI in a blog post, he also co-signed a document warning about its potential to cause harm.
Microsoft’s chief scientific officer, Eric Horvitz, shared his belief in the ability of artificial intelligence to enhance human flourishing. However, he appeared to have reservations as well.
In his blog post titled “Reflections on AI and the future of human flourishing,” published on May 30, Horvitz discussed the future of AI and highlighted a series of essays written by AI experts who had early access to OpenAI’s GPT-4 before its public launch.
Horvitz acknowledged being surprised by the capabilities of GPT-4 during his access in 2022, noting a significant advancement compared to its predecessor, GPT-3.5. He emphasized GPT-4’s ability to interpret intentions and provide sophisticated responses to various prompts, likening it to a phase transition, reminiscent of emergent phenomena observed in physics.
While testing GPT-4, Horvitz increasingly recognized the tremendous potential for transformative impact held by the model and its successors.
Both Microsoft and OpenAI provided early access to GPT-4 to a group of experts and invited them to write essays, aiming to explore the opportunities and challenges posed by the technology in various industries.
Simultaneously, on the same day as the announcement blog post, Horvitz’s name appeared as a signatory on a document published by the Center for AI Safety.
The document, endorsed by a wide range of AI experts, including Horvitz, called for prioritizing the mitigation of AI-related risks to prevent potential extinction, similar to global priorities given to risks such as pandemics and nuclear war. However, there were differing opinions on the significance of undefined risks related to future AI systems, with some experts elaborating on their reasons for signing and others expressing concerns about the potential distraction from more pressing issues.
In his blog post, Horvitz did acknowledge the challenges of developing safe AI and expressed anxiety regarding the transformative nature of the technology. However, there was no mention of potential extinction.
Apart from AI, Microsoft is also exploring other emerging technologies. Recently, the company was approved to participate in a central bank digital currency pilot project for the central bank of Brazil, alongside Santander, Visa, and several other applicants.