Good morning. Late tonight — 10 p.m. EST — Elon Musk will unveil Tesla’s robotaxi. You can watch the event unfold live here.

We’ll be exploring the event on Monday, but, in advance, let me just say that Mr. Musk has been predicting fleets of robotaxis for nearly a decade. All he has delivered on that front thus far has been hype … we’ll see if things are different this time.

I’m not gonna hold my breath.

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🌎 AI for Good: Hurricane prediction 

• 🏛️ DOJ considers a Google breakup

• 📊 TSMC is beating revenue expectations 

• 🧠 The Nobel Prize in Chemistry goes to … AI

AI for Good: Hurricane prediction 

With Hurricane Milton hitting Florida today, I thought it would make sense to take a look at the work that the National Oceanic and Atmospheric Administration (NOAA) has done in hurricane modeling and prediction. 

While the models that NOAA used to predict Milton’s path — and the far-reaching intensity of the storm — are traditional, physics-based models, researchers are increasingly looking to machine learning and AI as the next generation of storm prediction. 

The Hurricane Analysis and Forecast System (HAFS) refers to NOAA’s latest storm predictor; it is a numerical model and “data assimilation” system that is more reliable than previous models. 

• The model provides more accurate guidance on storm track, intensity and structure, including accurate predictions regarding sudden changes in a given storm. 

• A big update with this model was the ability to forecast storms seven days in advance, which gives people time to prepare, or in the case of Milton, to evacuate. 

Milton intensified Monday morning into a Category 5 hurricane expected to make landfall, thanks to hurricane modeling, late Wednesday night. Mandatory evacuation orders went out across Florida on Monday, two days before the storm hit. 

If you live in Florida, here are state government resources to deal with the storm. Be safe. 

Guidde - Create how-to video guides fast and easy with AI

Tired of explaining the same thing over and over again to your colleagues?

It’s time to delegate that work to AI. Guidde is a GPT-powered tool that helps you explain the most complex tasks in seconds with AI-generated documentation.

• Share or embed your guide anywhere

• Turn boring documentation into stunning visual guides

• Save valuable time by creating video documentation 11x faster

Simply click capture on the browser extension and the app will automatically generate step-by-step video guides complete with visuals, voiceover, and call to actions.

The best part? The extension is 100% free.

Check out Guidde

DOJ considers a Google breakup

Just a few months after a federal judge ruled that Google — in possession of 90% of the search market — holds a monopoly in search, the U.S. Department of Justice laid out a possible plan to remedy the monopoly. 

On the table, according to the DOJ’s 32-page court filing, are “behavioral and structural remedies that would prevent Google from using products such as Chrome, Play and Android to advantage Google search and Google search-related products and features — including emerging search access points and features, such as artificial intelligence — over rivals or new entrants.”

The AI of it all: Acknowledging that AI is not a substitute for search today, the DOJ said that the tech will “likely become an important feature of the evolving search industry.” 

• With this in mind, the DOJ said that it is “critical that any remedy carefully consider both past, present and emerging market realities to ensure that robust competition, not Google’s past monopolization, will govern the evolution of general search and text advertising,” suggestive of a far more expansive case than search alone. 

• Further acknowledging the role of data in search engines, the DOJ said it is considering requiring Google to make available to competitors “the indexes, data, feeds and models used for Google search, including those used in AI-assisted search features.”

Note: These things are far from decided. Judge Amit Mehta aims to rule on the recommendations by August of 2025. 

Google called the proposals “radical and sweeping.”

The Biggest Breakthrough in Next-Gen Tech Since iPhone

History is being made in the $2T global entertainment & media industry and you can get a piece by investing in Elf Labs — but only for a limited time.

With over 100 historic trademark victories, Elf Labs owns rights to some of the highest-grossing characters in history, including Cinderella, Snow White, Little Mermaid, and more. These icons have generated tens of billions in merchandise revenue alone, since their inception.

Now, Elf Labs is revolutionizing these characters with patented next-gen technology, including AR, VR and advanced compression algorithms for an unprecedented level of immersion. From virtual reality — without headsets — to AI-powered talking toys, this may be the biggest disruption to IP since Disney.

With two projects already funded, there’s limited space left. Become an Elf Labs shareholder today before the round closes on Oct. 30. Plus you can earn up to 40% bonus shares & other exclusive investor perks!

• Hearst, which owns dozens of newspaper and magazine brands, announced a partnership with OpenAI.

• Former Amazon exec Dave Clark launched Auger, a logistics software startup, with $100 million in Series A funding.

• The editors protecting Wikipedia from AI hoaxes (404 Media).

• Anthropic says its chatbot could alter its hiring plans (The Information).

• How AI revolutionized protein science, but didn’t end it (Quanta Magazine).

• TikTok’s use of casino-like virtual currency to allegedly exploit children faces scrutiny in DC lawsuit (CNBC).

• OpenAI sees continued attempts by threat actors to use its models for election influence (Reuters).

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

• Global Tech and AI Leader: EY, Philadelphia, PA

• Machine Learning Specialist: Technogen, Inc., Remote

• Martin: A tool for productivity and time management.

• Otto Engineer: A tool to summarize digital content.

TSMC beats revenue expectations  

Taiwan Semiconductor Manufacturing Co. (TSMC) reported September sales of $7.82 billion, up nearly 40% year-over-year. When you add in its July and August results, TSMC reported third-quarter revenues of around $23.58 billion, a roughly 36% year-over-year increase that came in above analyst — and TSMC’s own — expectations. 

• TSMC’s U.S.-traded shares were down slightly following the release of the report. 

• The company will release its full third-quarter results on Oct. 17. 

Why TSMC matters: Nvidia has been described as the picks and shovels in the AI gold rush; this position has sent Nvidia’s stock soaring since early 2023. While Nvidia supplies the AI firms, TSMC supplies Nvidia … strong numbers from TSMC indicate that demand remains strong, suggesting that there’s a bit of remaining health in the rather circular AI ecosystem.

The Nobel Prize in Chemistry goes to … AI

The computer scientists behind artificial intelligence models have been getting quite a bit of attention — in terms of Nobel Prizes — lately, beginning with Geoff Hinton and John Hopfield’s win of the Physics Nobel earlier this week. 

The trend continued Wednesday when three computer scientists — David Baker, John Jumper and Demis Hassabis (the latter two of Google DeepMind) — were jointly awarded the Nobel Prize in Chemistry for their work on computational biological problems in protein folding. 

Half of the $1.03 million cash prize will go to Baker, while Hassabis and Jumper will split the remaining half. 

• Baker, of the University of Washington, was honored for his work in the prediction and design of protein structures. He developed a software called Rosetta that, beginning in 2003, was able to design a completely novel protein structure. The software was later evolved — with deep learning at its heart — into RoseTTAFold. 

• The DeepMind scientists were honored for their work on AlphaFold 2, which has demonstrated an ability to predict the structures of nearly all known proteins based on their amino acid sequences. The model, released in 2020, seemingly solved the 50-year-old protein folding problem, which relates to three broad questions: 1, what is the code by which amino acid sequences dictate protein structures? 2, How can proteins fold so quickly? 3, Can computer algorithms be designed to predict protein structures based on their amino acid sequences? 

The impact: The impact of this work relates strongly to the core promise of AI in healthcare. Since proteins are the building blocks of life, a better understanding of how they form and interact with each other can allow scientists to create better medicines, individually targeted treatments and antibiotics that can overcome antibiotic resistance. 

It also opens up the field to address diseases that have thus far stymied modern medicine. 

Sans AI, the process of determining a protein’s structure was laborious, involving X-rays, flash freezes, high-powered microscopes and meticulous experimentation. The AI — developed in part by the three scientists honored by the Nobel Committee — has already sped this process up, enabling more targeted experimentation. 

But — by now, you should know that there’s always a but — as with all things in AI, there has been plenty of hype and misunderstanding around AlphaFold and its immediate impact. 

The biggest thing that is important to understand here is that AlphaFold predicts single protein structures. These predictions are not always accurate. Neither have these predictions solved biology or created revolutionary new drugs; what they’ve done is speed along the experimentation process, pointing scientists in a direction that then must be experimentally verified. 

• As one bioinformatics scientist wrote recently: “Alphafold is amazing for many applications but neither is it magic, nor does it replace actual experiments.” 

• “Alphafold is an amazing tool, but please please please, don’t think it’s omniscient and just magically knows what proteins look like.”

That’s not to mention the enormously long process of drug creation that involves years of testing, trials and studies necessarily preceding a public release, with AI or without it. 

The other problem is a little more complex. It begins with the fact that the prediction of protein structures — while a great advance — doesn’t tell scientists how or why protein folding happens, indicating that the protein folding problem remains unsolved. 

Princeton University computer scientist Ellen Zhong, who interned on the AlphaFold 2 team in 2021, told Quanta Magazine: “Right now, you just have this black box that can somehow tell you the folded states, but not actually how you get there.”

“For many people, you don’t need to know. They don’t care,” George Rose, a biophysics professor emeritus at Johns Hopkins, told the magazine. “But science, at least for the past 500 years or so … has been involved with trying to understand the process by which things occur.”

• Playing this out, in their native environments, proteins interact with surrounding molecules in binding processes that shift both their form and function.

• While DeepMind and Baker are working on predicting the structures of proteins interacting with other molecules (through AlphaFold 3 and RoseTTAFold All-Atom), the results aren’t as impactful as AlphaFold 2 was. 

AlphaFold 2 had plenty of training data through the Protein Data Bank, as Jumper himself has acknowledged. There’s far less training data available for the complexities of molecular compounds, something that could definitely hinder the revolution here. 

Computational biology is the promise of artificial intelligence. The notion of preventive, hyper-personalized healthcare will be life-and-society changing. 

But there is still tons of work here to do.

And AI is not some magic solution to the complexities of biochemical life.

🤔 Your thought process:

Selected Image 1 (TK):

• “Image one has some spiderwebs and other details (blemishes on the plants) that I don’t think AI would have added in such small amounts. Image #2 looks more natural in a sense but is too perfect.”

Selected Image 2 (Right):

• “The details on the waxy leaves (were) really stunning!”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Here’s your view on AI agents:

40% of you said they sound okay, so long as there is proper regulation. 24% are not interested in agents whatsoever, and 16% are super excited about them.

Regulate it:

• “Don’t all new ideas/processes need safeguards/guardrails?”