🤯 All of this week’s AI news

ALSO: ChatGPT Voice mode and Gemma 2 are out

Estimated Read Time: 7 to 8 minutes

We’ve got everything you need to know this week for the world of AI. Google’s a monopoly and announced Gemma 2. OpenAI released Voice mode while getting rid of more founders. All while Stability and Meta change video and asset creation forever.

• 🧑‍⚖️ Google is a monopoly, US Supreme Court rules.

• 📢 OpenAI’s Voice mode begins rollout.

• 🤖 Google releases Gemma 2 and 1.5 Pro matches GPT-4o.

• 🏃 How the Olympics use AI.

• 🎨 Stability releases upgraded 3D asset generator.

• 🫠 Another OpenAI founder leaves, new board member joins.

• 🎥 Meta reveals revolutionary AI video tracking.

🧑‍⚖️ Google is a monopoly, US Supreme Court rules

What happened: A U.S. district court judge ruled that Google violated antitrust laws by maintaining its search engine monopoly through “illegal” 👀 business practices. Here’s the breakdown and what it all means.

The breakdown: 

• Judge Amit P. Mehta ruled that Google violated Section 2 of the Sherman Antitrust Act by monopolizing the search engine market. Kinda late to the party though. It’s been like this awhile.

• Stats: By 2020, Google controlled nearly 90% of all internet searches and 95% of smartphone searches, with Bing as the closest competitor at just 6%.

• The Illegal Stuff: Google’s contracts with Android device manufacturers and Apple to make its search engine the default option were deemed illegal. It strengthens the monopoly. (Big Tech does the same in AI though.)

• 🥊 The judge highlighted that Google’s ability to degrade search quality without losing users is evidence of its monopoly. 😂

Now Amazon is in the crossfires too.

The UK is officially conducting an anti-trust investigation into Amazon’s relationship with AI company Anthropic, the makers of Claude 3 (one of the biggest AI models). They’re also investigating Microsoft’s relationship with OpenAI.

The 80/20: This ruling, if upheld, could lead to new competitors in the traditional search engine space.

But search is about to change forever with the advent of AI. SearchGPT is just the start. Combine this with the prevalence of younger generations to search through AI tools or social media channels (yes… like TikTok), and normal search was already on the way out. I wonder how much this ruling will really change in the end.

Convenient that Google gets to dominate traditional search for years, then when search is about to change up the Supreme Court finally rules them a monopoly.

🔥 Elon Musk sues Sam Altman again. Specifically he’s reviving the old lawsuit, which claims OpenAI puts profits above humanity’s good. Most of the suit focuses on OpenAI’s relationship with Microsoft and wants their relationship to become null and void.

📸 Runway releases image-to-video AI model. The model combines text prompts and a still image to create a very rudimentary video. It’s basic for now, but the tech is promising.

⚡️ Bing adds new AI features. The focus is on making search results more personal to the user. And considering Google’s Supreme Court troubles, this could be Bing’s big chance to take a share of the playing field.

🤦 Apple Intelligence will reportedly NOT launch with IOS 18’s initial release. These are just leaks and rumors so far. The big AI features will come to IOS later this year, but Mac users will gain access to the big AI features on time.

▶️ Youtube is testing new AI features to help creators brainstorm video ideas. The new “Brainstorm with Gemini” feature tries to come up with new video ideas, titles and even thumbnails. It’s a cool addition but I suspect it’ll be seriously flawed for a while.

📢 OpenAI’s Voice mode begins rollout

What happened: OpenAI has begun rolling out GPT-4o’s voice mode to a small number of GPT-Plus users (premium membership).

The details:

• The Selling Point: Voice mode can listen to you while it’s speaking and understand your tone of voice and emotional state.

• You’ll be able to share videos and your screen with GPT-4o, allowing the AI to help you with a document on your screen, a math problem, or whatever else.

• They sent out instructions to select GPT-Plus users. Check your email to see if you got in.

• OpenAI claims they’re concerned users will become too emotionally reliant on Voice mode. This sounds less like real concern and more like soft marketing to hype users and investors.

Why it matters: Voice mode’s tech demos are some of the most advanced uses for AI we’ve seen. The release will be a landmark moment if it can match the hype. And literally everyone will be using it.

🤖 Google releases Gemma 2 and 1.5 Pro matches GPT-4o

What happened: Google announced the Gemma 2 2B model, an upgrade to their super small, bite sized AI model capable of running on more hardware.

The details: 

• ⚡️ Gemma 2B runs on 2.6 billion parameters and was trained on a massive database of 2 trillion tokens.

• The model supposedly matches GPT-3.5 Turbo, despite being a super small, compact model.

• The model’s small size means it’s flexible for a variety of situations including local stations. Basically it consumes less power and can run on smaller amounts of hardware.

• Gemini Pro 1.5 has actually topped the AI leaderboard for performance.

Why this matters: Smaller, more efficient models are the future of AI in corporations. Running a massive data farm is expensive, whether you run one yourself or rent it out. Eliminating the need for such huge data centers, at least for most general AI models, is a massive point of improvement for AI as a whole.

🏃 How the Olympics use AI

What happened: The Olympics are using AI to conduct and edit the games. It’s a good indicator of how AI will be used across all sports outside of the Olympics soon. Here’s some of the ways the Olympics are using AI.

The Methods:

• The Olympics are using AI to censor/protect athletes on social media from harassment and emotional abuse.

• AI powered 3D tracking tech is studying athletes to provide greater insight to performance gains. Yes, robots are coaching the athletes (or at least trying to).

• AI is used in talent scouting, such as finding Olympic level athletes.

• TV broadcasts are using AI to create quick, personalized bites from the Olympics.

Takeaway: The Olympics are ahead of the game here, but what they’re doing will become the norm for every sport. AI will observe and analyze the athlete’s performance, coach them, and maybe even provide emotional support. We may even have AI powered announcers and commentators. It’s been possible for a while.

🎨 Stability releases upgraded 3D asset generator

What happened: Stability AI just revealed Stable Fast 3D, a pretty groundbreaking tool that converts single images into detailed 3D assets in just half a second.

The details:

• The Pitch: The model creates fully 3D models with full unwrapped meshes and textures built in. Basically it does all the hard work almost instantly.

• All the model needs is a single reference image, then it creates the 3D model in only 0.5 seconds on a GPU with 7GB VRAM.

• Optional re-meshing adds just 100-200 milliseconds to the processing time.

• Like all of Stability’s offerings, this model is free and available through Stability’s licensing arrangement, making it very powerful and influential.

Takeaway: Watch out 3D modelers! Soon it’ll be easier than ever for anyone to create 3D assets to make games, movies, or whatever else. The downside if that modelers lose job importance, though they’ll still be necessary to make adjustments and fine tune the models. If you think this is impressive, wait another few months. Something even better will come out.

🫠 Another OpenAI founder leaves, new board member joins

What happened: OpenAI lost another co-founder, John Schulman. He then immediately joined Anthropic (the makers of Claude 3). Well played, John.

The breakdown:

• John Schulman, co-founder of OpenAI, has left the company for competitor Anthropic.

• Meanwhile, OpenAI president Greg Brockman is taking a sabbatical, and product manager Peter Deng has also exited the company.

• His departure leaves only 3 of the original 11 OpenAI founders still at the company. It sort of looks like everyone is jumping ship. 🚢

• Schulman emphasized that his decision was personal and not due to a lack of support from OpenAI.

• In other news, OpenAI added Zico Kolter, a professor from Carnegie Mellon focusing on AI safety, to their Board of Directors. This feels like a distinct PR move to say, “Hey, we’re being safe!”

• Zico is probably Schulman’s replacement on the Safety Team.

The 80/20: Almost everyone connected to OpenAI’s original vision or safety teams leaves the company. This much drama, government interest, and structural change will eventually lead to a real blowout.

🎥 Meta reveals revolutionary AI video tracking

What happened: Meta announced their new video tracking AI model, Segment Anything Model 2 (SAM 2).

The details:

• SAM 2 can segment any object in an image or video, and consistently follow it across all frames of a video in real-time.

• Some examples are ping-pong balls in a real match, or the skateboard in a skating video.

• 💪 The big challenge with segmenting in videos was the changing nature of the objects. The object can change size, shape, or color. Meta claims SAM 2 has solved these challenges.

The 80/20: This has multiple uses. It could be used for special effects in films, sure, but it also helps identify people and objects for self driving cars and robots who have to identify objects in the real world. This is a desperately needed tech in the robotics world.

🎨 Ancient China!

Source: Midjourney

Give us your feedback!

Got feedback? Rate today’s newsletter by clicking below! 

Got more personal feedback? Fill in the box when you rate us and give us your criticism and feedback.

Thank you for reading! 

❤️Share the (ARTIFICIAL) Love!

Got a friend who wants to appear as smart as you? An AI fanatic or someone shaking in their boots of getting replaced? Share us by clicking the link below and you’ll get our love ;)