DeepMind Reminds Us Why We’re Willing to Flirt with AI Destruction

This is The Startup Breakdown, the newsletter where we learn, laugh, and love startups. By joining this growing community of hundreds of future startup aficionados (think i spelled that right?), you're getting a beachside view of the ocean that is the startup and VC scene. This ain’t your grandpa’s newsletter, so prepare yourself for an inbox full of 4/20 jokes and Succession references.

Happy Wednesday, folks.

Still awaiting entries for most chaotic Spotify Wrapped. Will accept entries through the end of the day. So far, the most unhinged Top 5 goes to the combination of:

1. Chief Keef

2. Taylor Swift

3. IZ

4. Katy Perry

5. Lana del Rey

I don’t know what you’re going through, but you’re either killing it or it’s doing a number on you. Whichever, stay strong, soldier.

Google DeepMind Reminds Us Why We’re Willing to Flirt with AI Destruction

Did a deep dive into the terrifying development that is Q*

And I had more than a few of you ask “If this could literally be the end of our existence, why are we investing so much time, money, and talent into it?”

The latest news out of Google’s DeepMind subsidiary answers that question, demonstrating that while the risks of the technology are genuine and terrifying (and genuinely terrifying), the potential breakthroughs which they enable make us a bit more willing to French kiss the Great White that is potential doom.

First, DeepMind has put on its hardhat and lab coat, waving its digital wand over Berkeley Lab’s Materials Project database and conjuring up 400,000 new materials in the process using the technology nicknamed “GNoME.”

In other words, pre-AI, discovering new materials would take months if not years, spanning putting together puzzles of elemental combos, designing experiments, testing them out, and validating the findings.

With GNoME (Graph Networks for Materials Exploration), this process can be done in a fraction of the time as the machines are able to run through limitless combinations of elements and virtually simulate their effectiveness for use in real world applications like semiconductors, solar panels, and even rare disease detection processes.

If you thought the number was staggering, the reality is that it actually unearthed 2.2 million new crystals… 380,000 was the number of those deemed stable enough to really test.

These same processes can be applied to literally any other field. Tell it to cure AIDs, design new farming techniques for developing nations with poor agricultural conditions, or design a new form of carbon capture which will allow us to avoid topping the 1.5ºC limit. The possibilities (and impact) and infinite.

Though probably less revolutionary (sorry, your idea of Uber for dog groomers won’t extend humanity’s lifetime, though I bet my dog Charli would look fabulous with a new blowout), these same techniques could be used for a startup, freeing up hours of time and resources and allowing for complex problems to be solved even by those without the book knowledge to properly attack them.

Hell, I taught myself to “code” by just getting good at telling ChatGPT what I wanted it to write for me 🤷 

So… that’s why we’re willing to flirt with AI-driven self-destruction.

Beyond the flash and dash of new materials and buzzwords, there's a greener pasture filled with sustainable technologies and solutions to the biggest problems facing humanities. Already, we’re making more progress in developing recyclable plastics, efficient batteries, and solar panels in a few weeks than we had in years prior.

For founders, not using AI to make your team more efficient and capable of solving complex problems which you might not have the know-how to tackle is akin to cutting off your own fingers and trying to race to write out Moby Dick while racing a computer that has the digital version and can simply copy and paste it.

DeepMind’s latest AI escapade isn’t just a cool science story; it’s a clarion call to all startups. It's a chance to leapfrog stages of R&D, cut costs, and hit the ground running with innovations that could define the next decade.

I know that I feature Venture Scout fairly regularly in my recommendations section, but it’s for good reason. I feel 10x smarter every time I open an email. Do your brain a favor and subscribe to this weekly newsletter today:

Cheers to another day,

Trey