The Signal: California tames AI’s wild west, microscopic superpowers, and retro game reborn

Hey friends 👋 Happy Sunday.

Here’s your weekly dose of AI and introspection.

Want to feature here? Sponsor this newsletter.

AI Highlights

1. California tames AI's wild west.

This week, California lawmakers approved a suite of bills aimed at regulating artificial intelligence, combating deepfakes, and protecting workers. The legislation covers everything from election integrity, explicit content, and worker rights for AI models.

Alex's take: With tech giants like Elon Musk and OpenAI throwing their weight behind these regulations. This unlikely alliance between Silicon Valley and lawmakers could just be the secret for responsible AI development. If California pulls this off, we might just see a blueprint for how to keep AI innovation booming while avoiding a Skynet situation.

2. AI's new superpower: microscopic sight.

The tool, AINU (AI of the NUcleus), can differentiate cancer cells from normal cells and detect viral infections at their earliest stages. Using super-resolution microscopy, AINU can analyse cellular structures 5,000 times smaller than a human hair.

Alex's take: This could completely change how we spot diseases before they even know they're there. While it's not ready for hospitals yet, I see tremendous opportunity for early detection and treatment.

3. Retro game DOOM reborn with AI.

Google's GameNGen AI system now simulates the classic game DOOM in real-time, running at 20+ fps on a single chip. Trained on 900M frames, it creates 3-second clips nearly indistinguishable from the original, all without a traditional game engine.

Alex's take: GameNGen is powered entirely by a neural model, specifically a diffusion model, rather than software algorithms and rules. From the player's perspective, the interaction is similar to a traditional game engine. They provide inputs and see the game respond in real time. However, GameNGen generates each frame as an image prediction rather than updating and rendering a symbolic game state. This could lead to dynamically generated, personalised gaming experiences in the future—how awesome is this?

1 Article I Enjoyed

Mark Zuckerberg and Daniel Ek on why Europe should embrace open-source AI.

Zuck and Ek are advocating for open-source AI in Europe.

Europe’s complex and inconsistent regulations are nothing new. Living in the UK, I’ve witnessed it firsthand.

They argue that ensuring power isn't concentrated among a few large players will help create a level playing field and allow European companies to drive forward scientific progress.

I think this is the right move: we’re at such a formative stage with AI development. We’ve got to remember that the “ChatGPT moment” happened less than 2 years ago.

I believe the risk lies not in AI replacing you but in a small group of individuals controlling the AI to control the narrative it feeds you.

We should fear people who think open is dangerous, when actually, open is the foundation for all greatness.

1 Idea I Learned

The future of dating isn't just swipes and likes—it's AI-powered.

Major dating apps like Tinder and Bumble are developing AI “wingmen” to help users craft better profiles, generate witty openers, and even provide real-time flirting feedback.

As AI becomes more integrated into our romantic lives, I feel we may need to redefine what we consider “authentic” in relationships.

Are we unlocking our inner charismatic selves or creating AI-driven personas?

Just because we can use AI to optimise our dating lives, should we?

What might we lose in the pursuit of algorithmic compatibility?

On one hand, AI could help people overcome social anxieties, express themselves better, and potentially find more compatible partners.

It could be seen as a tool to bring out the best in ourselves—or help us channel our inner Joey Tribbiani, if you will… How you doin’?

On the other hand, there's a risk of over-reliance on AI, potentially leading to a loss of genuine human connection.

If our interactions are constantly mediated by AI, we might lose the spontaneity and authenticity that often spark real chemistry between people—what I feel makes us most human.

Black Mirror (because I’m a big fan) parallels are particularly apt here. Like in “Hang the DJ,” we might find ourselves trusting algorithms more than our own instincts when it comes to compatibility. Or, as in "Striking Vipers," our romantic and sexual interactions could become increasingly virtual and detached from physical reality.

What constitutes a “real” connection in an AI-enhanced dating world?

The future of human relationships is already bridging the gap from sci-fi to sci-fact.

1 Quote to Share

“‘He was in mystic delirium’: was this hermit mathematician a forgotten genius whose ideas could transform AI – or a lonely madman?”

Alexander Grothendieck's revolutionary mathematical insights reshaped entire fields, but his relentless pursuit of truth led him to isolation and mystical obsessions.

As AI companies now explore Grothendieck's esoteric concepts like toposes, we're left wondering: could the scribbles of a reclusive hermit hold the key to next-generation artificial intelligence?

It's a reminder that groundbreaking ideas can come from unexpected places, and that genius often walks hand in hand with eccentricity.

Source: The Guardian

1 Question to Ponder

“I’m a total beginner with ChatGPT. Can you share a prompt you use a lot to help with your writing?”

Getting the right output from LLMs like ChatGPT can be a real pain.

“Expand this”, “Tweak that”… short commands only get you so far.

To really control the output of language models, you’ve got to add the right Style, understand the Audience, adjust the Length, and control the Tone.

Or, as I like to say, just add SALT.

Let’s look at some examples for how to actually apply it.

• Style: "List the steps to make blueberry pancakes" Result: Clear, organised instructions

• Audience: "Explain machine learning like I’m a 5-year-old" Result: Simple, engaging explanation

• Length: "Write a haiku about prompt engineering" Result: Concise, poetic summary

• Tone: "Explain quantum computing in a playful manner" Result: Fun, memorable explanation

This is a great prompting framework to help you move away from just asking basic questions.

If you provide the right guardrails, you’ll get the right response.

Start by adding SALT.

💡 If you enjoyed this issue, share it with a friend.

See you next week,

Alex Banks

P.S. Sorry if this keeps you up at night