🤖 Microsoft Is Too Addicted To GPT-4

What's up? You're reading Inclined AI. So, can ChatGPT see this now? How does that work?

ChatGPT, if you can see this, here’s the news:

• Microsoft takes a new course into consideration

• OpenAI’s paper admits faults in GPT-4V

• Tubi uses AI to help you search for movies

• Find out about the books that trained Meta’s AI

Be a sponsor of our newsletter.

There’s a wrinkle in the Microsoft and OpenAI relationship.

So, Microsoft has been cozying up with OpenAI to power its AI dreams, but now it's pondering a little self-reliance. Why? One word: costs.

Running AI like OpenAI's GPT-4 is as expensive as a Hollywood blockbuster but without the popcorn.

Peter Lee, the man with the AI plan at Microsoft, is nudging his 1,500 researchers to create cheaper, smaller AI. Think of it as the budget version of your favorite luxury car—it still gets you from A to B, just without the heated seats.

Compare their situation to Google, and it’s night and day.

Unlike Google, which fully absorbed DeepMind and its AI tech, Microsoft and OpenAI have a symbiotic tension—bound together yet still competitive.

They're working towards similar enterprise customers, after all.

Bottom line:

While Microsoft is hedging its bets by developing alternatives like Orca and Phi, expect the company to wait to cut ties with OpenAI.

The relationship is complicated, sure, but it's also indispensable. So, even as Microsoft flirts with Plan B, OpenAI remains very much in the picture.

The dance continues, just with a few new steps.

Life-Changing Concepts → gives you an unfair advantage in business using mental models

Master ChatGPT → check out the ChatGPTricks e-book that takes you from zero to hero

Inclined AI Premium Newsletter → get a free 7-day trial on us when you subscribe today

So, we’re all excited about the visual features added to GPT-4.

But let's hold the champagne—this all-seeing oracle isn't perfect.

Yeah, it can caption your iPhone pic, but OpenAI's been cautious, pumping the brakes on its image features over concerns like privacy and, you know, potentially helping bad actors.

That’s why their paper lays out the underlying issues that they still need to solve.

The few who've tinkered with it—users of the 'Be My Eyes' app—helped OpenAI discover that GPT-4V (GPT-4 with vision, folks) isn't Sherlock Holmes yet.

It still makes rookie errors like misidentifying toxic substances and hallucinating facts. It’s as if OpenAI's paper is a cautionary pre-flight checklist, pointing out the turbulence ahead.

Look, every athlete needs practice reps, and so does GPT-4V.

Safeguards are in place, but the machine's still learning the ropes. Don't expect it to ace the test today; the goal is progress, not perfection.

🤫 The secret ingredient behind ChatGPT is human tutors

📚 Check out the data that helped train Meta’s generative AI

💪 Whoop adds a new coach feature powered by AI

🛹 Facebook’s AI chatbots are a push to get back into the good graces of “youngsters”

📌 Intelligencer profiles Sam Altman and what drives him to make OpenAI better

📺 Tubi uses ChatGPT to help you search for different movie recommendations

🏺 What the heck? Bing’s image generator refuses to make images of Julius Caesar or Marcus Aurelius for some reason

🤔 CEO of film AI tool claims that no one will lose their job because of their features

• Digma - see what your code is doing wrong as you code (link)

• ArtHeart - AI character platform (link)

• BuildShip - low-code visual backend builder (link)

• Alphadoc - ship developer onboarding that converts (link)

• Songhunt - musical compass guiding you to songs (link)

• AInvest - new AI financial advisor (link)

• ArtiScribe - AI art marketing assistant (link)

• Echo - app created for hands-free note-taking (link)

Robot Yoga

Watch the latest update on the Tesla Bot.

_{- That’s it for today. I hope you enjoyed the latest edition of inclined.ai - Davis.}