Daily AI News Podcast · Alice Mallory and Bob Trent

Daily AI News Podcast: Wednesday December 18, 2024 - Anthropic Hacked & OpenAI o1 Now in API

·27 min·1 clip

It seems like there are new models every week from companies like OpenAI and Google, and governments are getting involved now.

The conversation starts with the pace of AI change. New models keep landing, and OpenAI and Google keep pushing the field forward, with governments getting involved too. The hosts quickly narrow in on OpenAI's O1 model. They treat it as a meaningful step toward better speed and accuracy. The focus stays on what the model can do in real situations. Function calling gets attention. Structured outputs do too. Image handling is part of the pitch as well. That makes the upgrade feel useful instead of abstract. The hosts then start asking what all of that means for everyday use. One example is medical diagnoses. Another is design work. They also point to code tasks that need visual understanding. Then the mood shifts. Hallucinations are still a real problem. Grounding comes up as helpful, but not as a magic fix. Reinforcement learning is one path. It rewards accurate answers and penalizes bad ones. External knowledge sources are another. Specialized models also enter the picture. Those are built for jobs where accuracy matters a lot, like scientific summaries or legal documents. Interpretability matters too. The hosts want more than a correct answer. They want a better sense of how the model got there. That helps with spotting bias and errors. The final stretch widens out to government involvement in the AI landscape, especially in the U.S. and the U.K. The episode closes with a familiar but important question: how do people shape AI responsibly while the field keeps moving this fast?

As heard by us

A brisk look at how AI is moving from novelty toward practical use.

The segment puts OpenAI's o1 in the foreground, but the real story is the push to make AI useful in more ordinary ways. It moves from speed and accuracy to function calling, structured outputs, and image handling without lingering too long on any one point, which keeps the…

Read the full review in PlayNext →

Why you'd press play

You want the practical meaning of OpenAI's o1 without wading through launch-day noise.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Daily AI News Podcast