From Possible to Probable AI Models
The article discusses the challenges of building reliable AI systems, emphasizing the distinction between what is possible and what is probable. It highlights that while generative AI can produce a wide range of outputs, the reliability of these outputs is often low. The author argues for a deeper understanding of probability theory to improve AI consistency in production environments.
- ▪Generative AI can produce a variety of outputs, but not all are reliable or useful.
- ▪The article emphasizes the difference between possible outcomes and probable outcomes in AI systems.
- ▪Hallucinations in AI occur when models generate plausible but low-probability outputs.
Opening excerpt (first ~120 words) tap to expand
Artificial Intelligence From Possible to Probable AI Models The real challenge in building reliable AI Sara A. Metwalli May 20, 2026 7 min read Share Image by Roberto Lee Cortes from Pexels. Over the past couple of years, I have been involved in many conversations about generative AI (and you probably have, too!). These conversations varied in focus, from ones with the general public about the use of AI to ones with more technical people about the accuracy of models. Regardless of who I am conversing with, people are often fascinated and curious about what models can do. Can an LLM write a functional kernel driver? It can. Can it write a song about how much you love your cat? It sure can. Can a diffusion model generate a photo-realistic image of a medieval astronaut? It can.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Towards Data Science.
Discussion
0 commentsMore from Towards Data Science
