(BNS) Simon Willison And SWYX Tell Us Where AI Is In 2025

Key takeaways

AI models have become cheaper and faster, with advancements in multimodal capabilities but have not significantly increased in intelligence since GPT-4.
AI agents face challenges in reliability and trust, which will need to be addressed for wider adoption in real-world applications.
Creative industries are adapting to AI's presence, raising questions about authenticity and the role of AI creators.
User interfaces for LLMs are in need of improvement to facilitate easier and more intuitive interactions.
The regulatory landscape around AI needs to evolve to address issues relating to usage, privacy, and societal impact.

Summary

The State of AI in 2025

In this segment, Brian, Simon, and Swyx discuss the advancements in AI leading into 2025. Simon notes that AI models have become more efficient, cheaper, and faster compared to 2024. Despite the improvements in multimodal capabilities, such as image and audio processing, there hasn't been a significant step forward in intelligence level beyond GPT-4. The expectation of a GPT-5 level model has been tempered by the reality of only price and capability enhancements.

AI Agents: Challenges and Opportunities

The conversation shifts to the concept of AI agents and their evolution. Simon expresses skepticism regarding their reliability, emphasizing the gullibility of language models. This leads to discussions about practical applications and limitations, contrasting the promise of automation with challenges in trust and safety. Both Simon and Swyx agree that while significant progress is observed, a deeper understanding and development of AI agents are still needed.

AI in Creative Industries

The podcast delves into how AI is impacting the creative sector. Challenges related to the acceptance of AI within creative fields are highlighted, especially regarding how creators can leverage AI tools for efficiency without losing authenticity. Discussions of AI influencers and creative autonomy spark interest about the future relationship between AI and artistry.

User Interfaces for LLMs and Local Models

The need for more intuitive user interfaces for interacting with language models is stressed. Both guests elaborate on the current complexities and UX hurdles in using LLMs, advocating for GUI solutions that simplify usage. The segment also discusses the shift towards local models, which have gained traction and improved capabilities, permitting more users to run advanced models directly on their machines.

Future Trends and Regulatory Landscape

As the podcast approaches its conclusion, Simon and Swyx examine the potential directions for AI development and regulation. With an eye on privacy, transparency, and accountability, they express the necessity for a robust framework that addresses AI's societal impact. This includes discussions on the challenges facing companies like OpenAI amid increasing competition and the evolving landscape around LLM discourse.

Show notes

The great Simon Willison joins SWYX and I to talk about everything we learned about LLMs in 2024, and what the state of AI is generally, as we go into 2025.
Here is Simon's blog post we keep referring to:
https://simonwillison.net/2024/Dec/31...
00:00 The State of AI in 2025
10:05 The Evolution of AI Models
19:54 Challenges in AI Agents
30:07 The Future of AI in Creative Industries
38:29 The Rise of AI Influencers
40:54 Credibility in the Age of AI
43:15 The Future of User Interfaces for LLMs
51:17 Local LLMs and Desktop AI Applications
55:17 AI Tools and Applications for Everyday Use
01:01:26 The Future of OpenAI and AI Regulation
01:08:08 The Need for Better Criticism of LLMs
01:10:41 The Future of Wearables and AI Integration