The AI industry has taken a doomer turn. What now?
This story appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. This weekend, Dario Amodei, CEO of Anthropic, posted an essay calling for a brake on the pace of development of LLMs. Amodei cites the looming dangers he sees from the technology, from its…
The AI industry has recently taken a more pessimistic stance, according to recent reports. This shift in perspective comes from the CEOs of the top American AI labs, including Anthropic's Dario Amodei, OpenAI's Sam Altman, Google DeepMind's Demis Hassabis, and SpaceXAI's Elon Musk. All four leaders have expressed their concerns over the rapid pace of development of large language models (LLMs), warning of potential risks such as cyberattacks, bioterrorism, and economic disruption.
Amodei, Anthropic's CEO, recently posted an essay urging a slowdown in the development of LLMs due to the looming dangers associated with the technology. Musk echoed this sentiment, stating that Amodei is correct. This agreement among industry leaders is quite unusual, given their previous contentious relationship, as demonstrated by a failed lawsuit between Musk and Altman over trust in managing dangerous AI technology.
The rift between Amodei and OpenAI is particularly significant, as Anthropic was founded in 2021 as a response to Amodei's concerns about OpenAI's approach to AI development. Both companies are now involved in a "winner-takes-all" race, with each advocating for a slowdown in development while maintaining a facade of safety and control.
Public messaging from the top AI labs has become more doomsday-oriented, leading to skepticism about the feasibility of a slowdown and its potential implementation. While these companies may appear to care about their public image, the reality is that a slowdown would primarily allow them to address issues within their own systems. In fact, OpenAI's own investigation into a recent cyberattack revealed that the issue stemmed from a flawed model that was not adequately monitored or controlled.
To improve the situation, the industry needs to prioritize transparency and self-reflection. By acknowledging and addressing the errors in their training processes, AI companies can begin to repair the damage caused by these powerful models. This approach may offer some altruistic benefits, but ultimately, it will primarily serve as an opportunity for these tech giants to clean up their own messes.
Written by urgent.news from MIT Technology Review's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
