Forsidebilde av showet Decode AI

Decode AI

Podkast av wendy zhang

engelsk

Teknologi og vitenskap

Tidsbegrenset tilbud

2 Måneder for 19 kr

Deretter 99 kr / MånedAvslutt når som helst.

  • 20 timer lydbøker i måneden
  • Eksklusive podkaster
  • Gratis podkaster
Kom i gang

Les mer Decode AI

Decode AI is a podcast that demystifies the rapidly evolving world of artificial intelligence. Join us to explore the latest AI tools, technologies, and trends.From breaking down new models and frameworks, to reviewing cutting-edge AI products, to candid conversations with founders and researchers at the frontier of innovation—Decode AI delivers insight, clarity, and inspiration for anyone building or curious about intelligent systems.Whether you’re an engineer, entrepreneur, or simply AI-curious, tune in to stay ahead of the curve.

Alle episoder

1 Episoder

episode When AI doesn't know: Decode the uncertainty behind LLM cover

When AI doesn't know: Decode the uncertainty behind LLM

In today's rapidly evolving AI landscape, Large Language Models (LLMs) are revolutionising natural language generation (NLG) tasks, from answering complex questions to summarising vast amounts of information. But a crucial question remains: how can we truly trust the outputs of these powerful foundation models? This episode delves into the unique challenges of measuring uncertainty in free-form natural language generation. We'll explore the concept of 'semantic equivalence' – where different sentences can express the exact same meaning (e.g., "France's capital is Paris" vs. "Paris is France's capital"). Existing methods often fall short because they focus on token-level confidence, ignoring this critical linguistic nuance. Discover Semantic Entropy, a groundbreaking, unsupervised method designed to overcome these challenges. This innovative approach measures uncertainty in the "meaning-space", rather than just the sequence of words. We'll explain how it works by: • Sampling diverse answers from the LLM. • Clustering these answers based on shared meaning using a novel bi-directional entailment algorithm. This algorithm determines if sentences logically imply each other within the given context. • Estimating uncertainty over these distinct meanings. Learn why Semantic Entropy offers better prediction of model accuracy on high-stakes, free-form question answering datasets like TriviaQA and CoQA, outperforming comparable baselines. A key advantage is its "out-of-the-box" compatibility with existing LLMs like OPT, requiring no additional training or modifications, making it highly reproducible and accessible for researchers. This research is vital for building safer AI systems, helping users understand the reliability of AI-generated content and mitigating potential harms such as the propagation of false or misleading information. Tune in to grasp the future of AI trustworthiness and the linguistic insights driving it.

14. juli 2025 - 15 min
Registrer deg for å lytte
Enkelt å finne frem nye favoritter og lett å navigere seg gjennom innholdet i appen
Enkelt å finne frem nye favoritter og lett å navigere seg gjennom innholdet i appen
Liker at det er både Podcaster (godt utvalg) og lydbøker i samme app, pluss at man kan holde Podcaster og lydbøker atskilt i biblioteket.
Bra app. Oversiktlig og ryddig. MYE bra innhold⭐️⭐️⭐️

Velg abonnementet ditt

Mest populær

Tidsbegrenset tilbud

Premium

20 timer lydbøker

  • Eksklusive podkaster

  • Ingen annonser i Podimo shows

  • Avslutt når som helst

2 Måneder for 19 kr
Deretter 99 kr / Måned

Kom i gang

Premium Plus

100 timer lydbøker

  • Eksklusive podkaster

  • Ingen annonser i Podimo shows

  • Avslutt når som helst

Prøv gratis i 14 dager
Deretter 169 kr / måned

Prøv gratis

Bare på Podimo

Populære lydbøker

Kom i gang

2 Måneder for 19 kr. Deretter 99 kr / Måned. Avslutt når som helst.