Ben Evans on AGI, DeepMind Ethics, Sakana, and China Deathbots

News and tools via the AI world.

Morning y’all!

Got a lot of rest and today should be better; less jet lag I hope! Short and sweet today.

Have a good one!

※\(^o^)/※

— Summer

---

Sakana released EvoSDXL-JP, a fast text-to-image generator tailored for Japanese-style images. Very neat!

Oh:

This model, the Image Joint Embedding Predictive Architecture (I-JEPA), learns by creating an internal model of the outside world, which compares abstract representations of images (rather than comparing the pixels themselves). I-JEPA delivers strong performance on multiple computer vision tasks, and it’s much more computationally efficient than other widely used computer vision models.

Meta continues to build cool stuff for others.

Straight from Google DeepMind comes a paper on AI ethics and assistants:

This paper focuses on the opportunities and the ethical and societal risks posed by advanced AI assistants. We define advanced AI assistants as artificial agents with natural language interfaces, whose function is to plan and execute sequences of actions on behalf of a user – across one or more domains – in line with the user’s expectations. The paper starts by considering the technology itself, providing an overview of AI assistants, their technical foundations and potential range of applications.

It then explores questions around AI value alignment, well-being, safety and malicious uses. Extending the circle of inquiry further, we next consider the relationship between advanced AI assistants and individual users in more detail, exploring topics such as manipulation and persuasion, anthropomorphism, appropriate relationships, trust and privacy.

With this analysis in place, we consider the deployment of advanced assistants at a societal scale, focusing on cooperation, equity and access, misinformation, economic impact, the environment and how best to evaluate advanced AI assistants. Finally, we conclude by providing a range of recommendations for researchers, developers, policymakers and public stakeholders.

It’s a bit of a read but one that might interest.

I guess the age of giant AI models is over? Others agree:

Nick Frosst, a cofounder at Cohere who previously worked on AI at Google, says Altman’s feeling that going bigger will not work indefinitely rings true. He, too, believes that progress on transformers, the type of machine learning model at the heart of GPT-4 and its rivals, lies beyond scaling. “There are lots of ways of making transformers way, way better and more useful, and lots of them don’t involve adding parameters to the model,” he says. Frosst says that new AI model designs, or architectures, and further tuning based on human feedback are promising directions that many researchers are already exploring.

I really enjoyed this interview and respect Ben’s opinions on a lot of tech. Here’s a summary for you:

Benedict Evans talks around the comparison of AI to previous technological advancements, such as the Graphical User Interface (GUI) revolution or the spreadsheet. Benedict Evans argues that while AI has the potential to be transformative, it's not yet clear how it will be applied in various industries or professions. The challenge lies in creating the necessary tooling and user interface to make AI accessible and useful for a wide range of users.

Ben also discusses the role of AI in businesses and the strategies companies should adopt, emphasizing the importance of understanding where in the stack these AI solutions sit and how they differ from traditional software or cloud infrastructure. The uncertainty surrounding the potential impact of artificial general intelligence (AGI) on society is also touched upon, with the speaker cautioning against assigning probabilities to unknowns and focusing on living one's life instead.

The conversation also covers the challenges of understanding and predicting the capabilities and limitations of large language models like ChatGPT and the importance of addressing bias in AI systems. Additionally, the current state and potential future of AR/VR technology, specifically the Apple Vision Pro and Meta's Quest devices, are discussed.

Overall, the video highlights the challenges and opportunities presented by AI and its potential impact on various industries and professions.

AI in the board room? That’s not that interesting but I think the idea about trusting these products to keep that information private and confidential is more interesting.

AI deathbots for folks in China:

Zhu, then 28, was shocked by how much the avatar of her late father was able to speak to her heart — for a moment last year, she felt like she was speaking to her dad again. “The experience made up for what I missed out with my dad,” Zhu recently told Rest of World. She hopes that advancements in AI technology would enable her late father to attend her wedding in hologram form. 

Grieving is a tough process.

Adobe Firefly has updated its tools. And it’s making me consider heading back to them and actually using them.

And that’s it. Simple one today folks.

※\(^o^)/※

— Summer