Good morning. In a move that feels very familiar to anyone who watched the Inflection - Microsoft saga unfold, Google has made a deal with Character.AI.

We break it all down for you below.

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

In today’s newsletter:

• 💉 AI for Good: ALS drug discovery

• 🤝 Google makes a deal with the founders of Character.AI

• 📈 Nvidia’s newest AI chip is delayed 3 months

• 📱 CivAI’s latest program can deepfake you in under 5 seconds

AI for Good: ALS drug discovery

ALS, known also as Lou Gehrig’s disease, doesn’t currently have a cure. 

Because of this, it has become an area of focus for some biotech and pharmaceutical companies. 

It’s an area that Verge Genomics, a biotech company built around an AI-powered drug discovery and development platform, has been making headway in. 

The details: After evaluating 11.4 million data points from ALS patient tissue and genetic datasets, Verge’s Converge platform discovered a new causative mechanism for ALS, and developed a new therapeutic target to address it. 

• This target was discovered and delivered into the clinic in just four years. 

• The drug has completed pre-clinical and phase 1 trials; it is currently undergoing a Proof of Concept study in Canada and several other European countries. 

Verge in March announced a strategic collaboration with pharmaceutical company Ferer to develop the drug, which Verge said has “shown efficacy” throughout its pre-clinical run of trials. 

Why it matters: “For years, researchers have heavily relied on animal or cell models to identify new targets, in a way that oversimplifies the enormous complexity of human biology, particularly for diseases like ALS,” Dr. Robert H. Scannevin, Verge’s chief scientific officer, said in a statement. “The Converge platform starts with, and integrates, human data and human model systems throughout discovery and development,” providing “unique insights into the biological underpinnings of ALS.”

10x Your Outbound With AI BDR, Ava

Imagine your calendar filling with qualified sales meetings, on autopilot. That's Ava's job. She's an AI BDR who automates your entire outbound demand generation. 

Ava operates within the Artisan platform, which consolidates every tool you need for outbound:

• 300M+ High-Quality B2B Prospects

• Automated Lead Enrichment With 10+ Data Sources Included

• Full Email Deliverability Management

• Personalization Waterfall using LinkedIn, Twitter, Web Scraping & More

Book a demo and supercharge your sales team today.

Google makes a deal with the founders of Character.AI 

In yet another Big Tech/startup shakeup, Noam Shazeer — who led the team of Google researchers that built LaMDA before leaving to start Character.AI in 2021 — is rejoining the tech giant. 

• Note: Shazeer was one of the co-authors of the famous 2017 paper “Attention is All You Need,” which introduced Transformer architecture. 

The move is part of a wider deal announced Friday between Google and Character.AI; as part of the agreement, Character will provide Google with a “non-exclusive license for its current LLM technology.” In return, Character will receive “increased funding,” though it didn’t specify the amount. 

• As part of the deal, Shazeer and Daniel De Freitas, another Character.AI co-founder, will be joining Google DeepMind. 

• 30 members of Character.AI’s research team, according to The Information, will be joining them. 

• Character investors are being bought out at $88 per share, according to The Information, and employees are getting paid cash to match their stock, according to The Verge. No equity is changing hands. 

Some context: The situation seems awfully similar to Microsoft’s soft-acquisition of Inflection AI in March, in which Microsoft hired Inflection’s co-founder Mustafa Suleyman and the majority of the company’s staff in a $650 million deal. 

Both the U.S. Federal Trade Commission and the U.K. Competition and Markets Authority have said that they are investigating the deal.

Cognitive scientist and AI researcher Gary Marcus said that this pattern of well-funded startup founders jumping to Big Tech indicates that things are “starting to fizzle.” 

“2023 was the year of AI promise,” he said. “2024 is the year of AI reality.”

• Startup Contextual AI raised $80 million in Series A funding.

• AI chipmaker Cerebras confidentially filed for an IPO.

• DOJ sues TikTok for ‘Flagrantly Violating Children’s Privacy Law’ (FTC)

• For the first time, wind and solar generated more of the EU’s electricity than fossil fuels in the first half of this year (Euronews).

• US launches antitrust probe into Nvidia (The Information).

• Tech stocks see steepest three-week slump in two years, led by plunge in Amazon and Intel (CNBC).

• Meta Is Offering Hollywood Stars Millions for AI Voice Projects (Bloomberg).

Nvidia’s newest AI chip is delayed 

Nvidia’s upcoming Blackwell family of GPUs will be delayed by at least three months due to design flaws, according to The Information. 

The delay comes as a series of Big Tech corporations — including Meta, Tesla, Google, OpenAI and Microsoft — have ordered billions of dollars worth of Nvidia GPUs. The delay also coincides with a retreat of Nvidia’s stock, which has fallen some 15% in the past 30 days. 

The details: Nvidia unveiled its latest GPU — the Blackwell — in March; in May, the company said it planned to ship the chips later this year. Now, the chips likely won’t start shipping until the first quarter of 2025. 

• The delay could in turn delay the supercomputing plans of many Nvidia customers, something that could further test investors’ thinning patience to see some sort of return on massive investments in AI. 

• The Information noted that both Google and Meta have placed orders for Blackwell chips likely worth more than $10 billion, while Microsoft has increased the size of its order by 20%. 

"As we've stated before, Hopper demand is very strong, broad Blackwell sampling has started and production is on track to ramp in the second half," an Nvidia spokesperson said. 

Everyone tells you to learn AI but no one tells you where.

We have partnered with Growthschool to bring this ChatGTP & AI Workshop directly to you. (It normally costs $199, but readers of TDV get it free) 🎁

Register here for free – valid for next 24 hours only!

This workshop has been taken by 1 Million people across the globe, who have been able to:

• Build a business that makes $10,000 by just using AI tools

• Make quick & smarter decisions using AI-led data insights

• Write emails, content & more in seconds using AI

• Solve complex problems, research 10x faster & save 16 hours every week

You’ll wish you knew about this FREE AI Training sooner (Btw, it’s rated at 9.8/10 ⭐)

Save your seat for $0 now! (Valid for 100 people only)

CivAI’s latest program can deepfake you in under 5 seconds 

One of the most significant threats of generative artificial intelligence involves deepfakes. The severity of the threat here has to do with the breadth of the implications; deepfakes can be — and have already been — used to spread various types of misinformation, including the political and election-related kind; deepfaked images have impacted the stock market, supercharged fraud, phishing, thievery and identity theft and created a whole new way to harass women and girls online (through explicit and nonconsensual deepfake images). 

The world that deepfakes are ushering in is one of a stark, yet convincing, unreality; a digital environment that — by default — shouldn’t be trusted by its users. 

A non-profit called CivAI — co-founded by software engineers Lucas Hansen and Siddharth Hiregowdara — has made it its mission to help local governments better understand the threats at hand. 

• Earlier this year, they put me through their full-scale demo, designed to give government officials a more tangible understanding of the capabilities of off-the-shelf genAI technology. 

• Last week, they launched a new demo, one that, unlike the first one, is fully available to the public. This demo turns a selfie into a deepfake in less than five seconds. 

You can try the demo out here. 

“It's a way for someone to build a small amount of intuition around how AI actually works, to feel at a gut level that this is powerful and magical in novel ways,” Hiregowdara told me. “That’s the intuition that undergirds all of our work … we want to give people a concrete understanding by interacting with AI firsthand.”

• The demo is built entirely using off-the-shelf models and is loaded up with safety measures, including strict moderation layers and prompt limitations. 

And while deepfakes aren’t new, the point the founders want to drive home is that the proliferation of genAI has made it remarkably easy to scale convincing synthetic content; as Hiregowdara said: “The bar has gotten so much lower that the quantity has a quality all of its own.”

Right now, they said, we’re at something of a crossroads when it comes to deepfakes and AI in general; we can adapt regulation reactively and over time, or we can adopt regulation proactively to mitigate harms in advance.  

• The speed at which things are moving complicates regulation, but Hansen said it’s important not “to equate that sense of helplessness with the unmoderated approach being ideal.”

• Regulation, he said, is important, as are certain technical solutions that could assert content provenance and authenticity. The issue with the latter is enforcing their utilization across a variety of social media and telecom platforms. 

“But there isn't a mechanism to perhaps nudge (regulation) to happen,” Hansen said. “I think public opinion is one and that has been somewhat helpful. I think that's part of why education is important.”

The (brief) state of regulation: While the European Union’s AI Act recently began entering into force, U.S. AI legislation at the federal level remains largely nonexistent. The Defiance Act — which would allow victims to sue over the creation/sharing of explicit, nonconsensual deepfake images — recently passed the Senate. States, meanwhile, have begun to erect a patchwork layer of regulation boasting varying levels of strength. 

This all comes as AI companies continue to increase their lobbying efforts.

A poll before you go

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

We’ll see you in the next one.

Your view on AI search:

Around 20% of you said you don’t really need AI search; you’re perfectly happy with Google as-is. But 63% of you said you’re super down to explore AI search.

Been using it for research:

• “I have been using ChatGPT just to find how to do things, and so far (it) has been a great tool. I’m looking forward to this new application.”

Also using it for research:

• “I already use several different LLMs for search. Vastly better than Google except for real-time information.”