Hi, it's Tech editor Alexei Oreskovic filling in for David today.
It's a sign of the times—in the last few weeks, my schedule has been filled with AI-focused conferences, dinners, and events. TED had an AI-focused conference here in San Francisco last week, featuring speakers such as DeepMind cofounder Shane Legg, OpenAI cofounder Ilya Sutskever, and LinkedIn cofounder Reid Hoffman. A week earlier, Seattle-based VC firm Madrona hosted its second annual Intelligent Applications Summit.
One of the big takeaways that I've heard again and again at all these events is how quickly things are moving. Consider that when Madrona had its first IA Summit in October 2022, ChatGPT hadn't been released yet, and the event's agenda was about different machine learning trends. This year, it was all about generative AI.
This breakneck pace of change is exciting, but it's also a bit scary—whether you're a company trying to figure out how to implement AI within your organization or your products, an investor wondering how to invest wisely on the technology trend, or a policymaker thinking of how to regulate AI.
As recently as nine months ago, many people were looking at Open AI's GPT LLM as the main model. But today, said AWS SVP Matt Garman at the IA Summit, "You're probably hard-pressed to find anybody that thinks there's going to be one model."
Mixtures, or "ensembles," of different models are increasingly coming into play. But because the models themselves are constantly evolving, businesses building with AI need to continuously reappraise the individual models they've chosen to make sure each one is still the most appropriate.
One of the most memorable quotes that stuck in my mind came from Altimeter founder Brad Gerstner, who described the explosion of AI startups, many of them with massive valuations. Two "simultaneous truths" are in tension right now:
"One," Gerstner said, "is that this is going to be the biggest thing of all of our lives. And on the second hand, 90% of this stuff won't be here in five years."
We (hopefully) will be here in five years, even as AI models, AI companies, and AI use cases continue to evolve and morph. AI is moving fast, whether we like it or not. Are we prepared to make sure it doesn't break things?
With that, here's today's tech news.
Alexei Oreskovic
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
Today’s edition was curated by Rachyl Jones.
