- Inclined.ai
- Posts
- 🤖 Microsoft Germany Teases GPT-4
What's up? You're reading Inclined AI. Happy Friday! I’m going to spend the weekend at the movies watching Scream VI. Wish me luck.
Here's what you should watch/read:
Microsoft Germany might’ve dropped some big GPT-4 news
Noam Chomsky argues that ChatGPT isn’t all that great
Discord and Grammarly add new AI tools
OpenAI’s President responds to Elon Musk
THE GERMANS LET THE CAT OUT OF THE BAG
Prompt: a cat runs out of the bag Style: Gustav Klimt
Or did they? The CTO of Microsoft Germany confirmed GPT-4 would launch next week, according to Heise Online.
It sounds like a weird fever dream.
Heise Gruppe is a real media company, and this is too odd to be a lie, right? I studied German in high school. They’re not known for pranks.
What if this story is true? We’re in for one of the most hyped moments in AI this year. Football fans climbing greased light poles will look like child’s play compared to this celebration.
The reveal happened at “AI in Focus,” an information event hosted by Microsoft. The details are sparse, sadly. Even if I had a video from the event, I haven’t spoken German in almost a decade.
One word stood out though: multimodal.
Andreas Braun, the CTO who spilled the beans, indicated that the new model would be a Multimodal Language Model. He says that should be evident given the recently published paper about Kosmos-1, Microsoft’s MLLM.
It’s weird. Sam Altman said the exact opposite when confronted with this question at a leaked private Q&A.
So which is it: LLM or MLLM?
Here’s my advice, don’t bet any money on this, but keep your fingers crossed. We expect GPT-4 this quarter, but confirming the release does not matter as much as the product details.
We’re still lacking in that department.
A SPONSORED MOMENT
Are you tired of losing track of your favorite book quotes or struggling to keep your reading notes organized? Look no further than CommonPlace, the ultimate reading app for book lovers.
With CommonPlace, you can easily highlight and save your favorite quotes and ideas directly to your phone, making it simple to revisit them anytime, anywhere.
Say goodbye to the hassle of pen and paper and hello to a modern way of organizing your reading notes. CommonPlace is perfect for students, avid readers, and anyone looking to streamline their reading process.
Check out CommonPlace today and discover the joy of organized reading!
NOAM CHOMSKY DOESN’T BUY THE CHATGPT HYPE
The Linguistic Professor and public intellectual wrote a New York Times Op-Ed with some colleagues. The headline: ChatGPT is not the right step forward.
Some people hear Noam Chomsky, and their eyes instantly glaze over. So I’ll do my best to keep this light but informative.
To do that, I’ll highlight his big points and provide support and a counterpoint.
1/ We will not reach AGI if we let models like ChatGPT dominate the field
Support: These models don’t reflect authentic human thought patterns, and if we dive too deep into developing them, we end up steering off course.
Counter: No one said ChatGPT is the end solution. It’s a massive public leap forward that indicates the development of AI has the potential to advance rapidly. No one should expect ChatGPT to be the final word.
2/ ChatGPT’s brute force pattern-matching is nothing like how a toddler learns to speak and think
Support: ChatGPT and a child can understand that if you drop an apple, it will fall to the ground. That’s a description and prediction example of intelligence.
But ChatGPT lacks explanation and critical thinking in its model. As a result, it cannot explain what is not the case or what could be the case.
It only sees patterns and describes them.
Counter: I asked, “I have an apple in my hand. I drop it. What happens?”
ChatGPT responded, “When you let go of the apple, it will fall to the ground. This is because of the force of gravity, which is the force that attracts two objects with mass towards each other.”
Chomsky said that even if you get that answer it’s pseudoscience, so idk.
3/ True intelligence is marked by moral thinking. OpenAI prohibits ChatGPT from thinking this way.
Support: He’s correct. There is a rough-and-tumble nature to talking to ChatGPT.
Anytime you get that response, “As an AI model…,” that’s what he is talking about here.
Counter: Balancing between pleasing most users and avoiding objectionable content is a tall task for any model.
OpenAI is not ignorant of this criticism; every research team tackles this when crafting their models. We get closer to a solution with every paper.
There you have it.
A full rundown of Noam Chomsky’s Op-Ed made simple so that you can talk about it at your next dinner party.
Just don’t be this guy:
Quick Nuggets
🎉 Big Party: everyone in generative AI seems to be RSVP’ing yes to productivity apps
♻️ AI’s carbon footprint is growing with its user base
💰 An AI chip Phoenix rises from the ashes—its name: Mythic
🦙 LLaMA leaked, and Meta is left wondering how it happened
☎️ Call your parents but don’t use AI to replicate your voice like this CNN reporter
🤔 Oops: an AI ad spot by an LSU gymnast brought up ethics issues in college NIL deals
🪖 M*A*S*H stars agreed to do a table read of an episode written entirely by AI
💸 An empty account: Sam Altman reportedly poured a lot of his money into getting two of his companies off the ground
🍎 Smarter than a 5th grader? Not for long. This classroom is all-in on AI learning
🤝 OpenAI President responds to Elon Musk: “We made a mistake.”
📈 Prompt marketplaces are popping up left and right. What are they, and should you care?
🕶️ Visual ChatGPT is an open-source tool that was made to send images within chat AI
🦅 AI regulation is top of mind for the US Chamber of Commerce
🧠 Big Tech, Big Brains: the race to make AI brains for humanoid robots
🇨🇳 Chinese AI groups are using cloud services to evade export controls on US chips
🗞️ AI-assisted articles are helping one UK newspaper fill blank space
🚪 OpenXLA opens its doors to all researchers as part of Google’s Open Source initiative
⚡️ Chatbots are causing a mad scramble as internet businesses try to stay ahead
💭 ThoughtSpot adds a GPT-3 integration to assist in data queries for customers
⛑️ Healthcare costs could be cut down through the use of AI
📢 PR prompts are happening so often that Stagwell is responding with its own AI tool
💯 100 million active users: Bing crossed a big milestone in its race to retake some of the search engine market
💼 Your office job is not safe from AI. WIRED explains why that is a good thing.
📸 GigaGAN is the ultimate name for the ultimate generative network
🔥 Fresh Products
GrammarlyGO - is the company’s new AI-integrated assistant (link)
Discord - announced new AI tooling for the community (link)
Wonder - creates studio-grade CGI that automatically animates, lights, and composes CG characters (link)
WatchGPT - a chat AI service for your Apple Watch (link)
anywebsite.ai - turn any site you want into one with an integrated chatbot (link)
Replix - helps with content generation to amplify your brand (link)
SocialBee - added new AI features to help make and schedule content (link)
Type - write incredibly fast with this document editor (link)
Depth - use AI to explore how you are feeling (link)
Ai LiLi - bought an NFT? talk to it on Discord (link)
Sparkle - focus your cover letter on each application without rewriting it every time (link)
Gifideas AI - find the right gift for anyone (link)
Notability - send your notes, and it will organize them in Notion (link)
Adrenaline -get help debugging your code…or the code ChatGPT wrote (link)
Good Content, Snoop Mario
One, two, three, and to the fo'
Snoop Doggy Dogg and Dr. Mario is at the do'
That’s it for today. I hope you enjoyed the latest edition of inclined.ai - Davis.
Want more?
Upgrade today for less than the price of a latte a month and get benefits like:
A weekly Sunday deep dive into AI trends
Access to Q&A privileges to get your biggest AI questions answered
Exclusive discounts & invites to beta AI tools
