🤖Google makes their AI announcements

ALSO: Sutskever leaves OpenAI

Estimated Read Time: 6 minutes

Google is one of the biggest players in the AI space. And they just finished their I/O conference, featuring a ton of huge AI features coming our way. Today we’re covering everything they announced plus some OpenAI drama (as always🍿). Get ready… there’s a lot.

(Today’s letter almost exclusively focuses on these announcements.)

• 📲AI is coming to Google search.

• 📸New image and video generation AI.

• 🤖All of Gemini’s upcoming features.

• 🤔Ilya Sutskever leaves OpenAI amid drama.

🤖All of Google’s AI announcements

What happened: Google just had their big I/O keynote where they announced a ton of AI plans. Strap yourself in… there’s a lot to cover. There’s a plethora of Gemini-specific announcements too, so check them out further below.

The details: 

• AI is coming to Google search in an effort to fight off Perplexity and ChatGPT from taking market share. There’s also a lot of concern how AI will affect websites. Google says their AI will summarize searches when necessary, but stay out of the way when AI isn’t needed for the search. It’ll be best when the information is scattered or complex.

• Gemini will also reorganize search results depending on your search query (SEO gurus are going to have a fit). I like this idea because Google will otherwise show you the same useless links even when they aren’t what you’re looking for.

• AI in Google Search will also help with planning, such as organizing a trip or driving route. This stuff integrates with other apps like Google Maps.

• Google is releasing Imagen 3, their new image generation model. After the massive screw up with Gemini’s image generator, we can only hope the model isn’t as broken.

• Google’s also teasing Project Veo, a video generative model to compete with OpenAI’s Sora. They have a video here demonstrating its capabilities. Looks impressive!

• Google announced LearnLM, a generative AI model made specifically to help teach kids in school. The model is the product of a collabration between Deepmind and Google Research, and is designed to teach students “conversationally”. This obviously has massive potential to change the way we educate.

• Educational Youtube videos can have a “raise your hand” feature allowing you to ask an AI questions about the subject of the video. Then the AI will generate quizzes to test your knowledge. It’s honestly pretty cool sounding, but I have to wonder about potential inaccuracies. The feature will roll out to Android users first, so I gotta wait.🫠

• Android users will be able to “circle to search”, meaning they draw a circle around something on their screen, and AI will give them information about the object circled. Neat, but what’s really important is how this tech is applied to other areas in the future.

• They’re adding a new 27-billion-parameter model to Gemma 2. This has been a major request from developers. According to Google, it can run efficiently on a single TPU host and vertex AI. Wish granted developers!

• Android phones will automatically detect scam callers by using a gen-AI called Gemini Nano. One way they exampled is detecting when the caller is trying to retrieve passwords from the user. These tactics are pretty well known, but they prey on naive (and often old) people. So an AI assistant protecting naive people from scammers sounds good in my books. Please work well. I want this so bad.

• Firebase is adding Firebase Genkit to their platform to help developers create AI applications.

• Google their next generation of Tensor Processing Units (TPU) AI chips called Trillium. They release later this year. They have a 4.7x performance increase over the fifth generation and use SparseCore, a “specialized accelerator for processing ultra-large embeddings common in advanced ranking and recommendation workloads.” That’s a lot of lingo.

• They also announced new Pixel phones and tablets.

Check out the section further below for Gemini-specific announcements!

💣Every Gemini Announcement from Google

Gemini Everything Else: All of the Gemini (their big chatbot) news today.

• Gemini will enter into Gmail to help summarize and sort your emails. Basically a fully fledged email AI-assistant. However they included more complex features too: such as helping you process a return by first searching your inbox, then finding the receipt and finally filling out the online form. Some serious quality of life there, but something tells me it won’t run that smoothly in real life.

• They showed off Gemini 1.5 Pro some more, boasting huge context lengths and taking up to 2 million tokens. But nothing new per say. Select users (like myself 😌) can already use Gemini 1.5 Pro right now.

• Gemini Live is a feature allowing you to have live voice chats with Gemini. It’s kinda like… GPT-4o. Actually it’s exactly like that. Apparently it’ll adapt to your speech patterns in real time AND you can interrupt it while it’s talking (finally someone who doesn’t mind me interrupting them). Google claims this is a big advance in tech even if it doesn’t seem special on the outside.

• Gemini Nano, the smallest Gemini model, will be built directly into Chrome on desktop. Your own device will power the AI. The main example they used was helping you write. Hopefully it can do a lot more than that. It’s coming in Chrome 126.

• Gemini is coming to Android to replace Google Assistant. Many of the features we spoke of above will be built into Android, such as the ability to ask a Youtube video about specific info in the video.

• Google Maps is also getting Gemini, which will use AI to generate descriptions of places and events. Developers won’t have to enter their own descriptions anymore. I’m not sure how accurate this would be, yet alone if it’s that useful. But there it is.

Why it matters: Google is perhaps the biggest player in AI aside from Microsoft. What they do will affect how we all use technology. OpenAI wowed people by offering many features in a single place. Google seems like they’re trying to consolidate the tech into a single place. All of the AI features on Chrome, Gmail, Maps, or Android. That’s the next big step to wide-scale AI adoption.

🤔Ilya Sutskever is leaving OpenAI

What happened: Ilya Sutskever, chief scientist and co-founder of OpenAI, is leaving the company of his own volition.

The details: 

• Sutskever hasn’t said what he’ll do next.

• Greg Brockman and Sam Altman confirmed his leaving, saying he was essential to creating the company and pivotal to AI’s current development.

• GPT-4o’s release is sort of a send off to Sutskever’s legacy on the company. However, he was against much of the current direction OpenAI is headed.

• Jakub Pachocki, OpenAI’s director of research, will replace Sutkever. He joined the company around 2017, so he’s quite capable.

• Ilya was one of the founders of OpenAI. He’s been there since the beginning.

The REAL drama they don’t talk about 👀:

• It’s rumored Ilya and Sam Altman had a power struggle for the company, with Altman seeking to take the company in a more dangerous direction (profit, dangerous AI).

• Sam Altman was fired for a week last year after the board discovered he was creating something dangerous (possibly Q*). It’s believed Ilya had something to do with his sudden firing. Not even other OpenAI employees knew what was happening.

• Much of the old board, aside from Sutskever, resigned after Sam Altman rejoined. Obviously there’s some disagreement here.

• A while back two AI researchers, one of which had strong connections to Ilya, were fired for leaking info to the board. There’s a possibility these were the researchers who got Altman, who recently rejoined the board and became the de-facto leader of OpenAI again, fired last year.

• So yes… there’s a possibility Ilya left because of his conflict with Altman. There’s probably more we don’t know about. But it is fun to speculate. 🧐

Why this matters: This could be the turning point for OpenAI. Sam Altman has complete power now… and his vision will become OpenAI’s vision. Let’s hope it’s beneficial for humanity and not just progress for the sake of progress.

🌙 A logo you aren’t supposed to make…

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 ;)