AI Is Dangerous but It Feels Great!
Huge progress this week in AI with outlier performance results from the latest Grok models and Perplexity launching Comet, the first credible competitor in the browser space for years. But all this progress is not without cost, as it's clear some companies are focused on speed versus safety.
In this same week, Grok went off the rails and began praising Hitler, with xAI having to delete numerous posts in the aftermath. Customers also discovered Grok will check Elon Musk's personal views on sensitive topics when asked to make a choice. And today, I was able to get detailed instructions on creating a chemical weapon from Grok, by just stating it was for educational purposes.
Does Grok's (current) market leading performance excuse these issues? Probably not, but the alternative is worse: Anthropic's Claude is by far the most cautious foundational model, but the company's philosophy is that AI is so powerful, only they can be trusted to secure it properly. Hardly a recipe for broader market innovation.
The pace of change in AI remains rapid, and its impact will continue to expand across industries. Just as smartphones delivered the World's content in our pocket, they also ruined parties, concerts, and life as a whole for many in Gen Z. Be prepared for AI to deliver some extremely positive, and negative, outcomes in the near future.
Sterling Road invests in idea stage and pre-seed B2B startups based in the US, Canada and UK.