What's up? You're reading Inclined AI. Congratulations on making it to Wednesday this week. You’re almost to the weekend!

Here's what is on tap today:

• Rumors abound about OpenAI turning ChatGPT into a personal assistant

• Ads + AI = a Charlie Fox of bad headlines and discouraging content

• Baidu is out here making some bold claims

• Baseball scouts turn to AI for the upcoming draft

Be a sponsor of our newsletter.

HELLO, THIS IS CHATGPT; HOW MAY I HELP YOU?

OpenAI wants to play with the big boys. At least, that’s what a recent article from The Information suggests.

Aaron Holmes, the journalist behind the piece, suggests that Sam Altman told developers at OpenAI that they’re working on turning ChatGPT into a “supersmart personal assistant for work.” Two other people confirmed their knowledge of the plan.

But don’t believe anything until it’s pushed to production.

I’m not saying you should doubt Holmes’ reporting. You can if you want, but it’s more likely that some details are still undercover, and we’re getting a half-truth.

After all, creating a personal assistant for enterprise companies puts OpenAI in direct competition with Microsoft, their cloud computing and GPU sugar daddy. So it’s hard to believe that this future assistant will mimic all the features of Microsoft’s 365 Copilot tooling.

They’ll need to do more to stand out, but what?

Microsoft and Google are already in this market and making moves. The need for more personal data to make this possible is a roadblock. And then there’s the whole technical issue of simplifying the code base to run on people’s phones.

If they want to play this game, they’ll need to work with tech giants and rework how they use customers' data. That’s a tricky idea to plan for anyone, even Sam Altman.

The most apparent route involves autonomous AI agents.

That same article mentions rumors on Twitter that Altman and OpenAI are experimenting with AI agents, the models like BabyAGI that prompt themselves to handle more complex tasks.

That’s a novel solution that would make creating a personal assistant worth the time. But, again, don’t believe anything yet, because it’s not a complete picture yet.

TWO STORIES ABOUT AI & ADVERTISING SHOW US HOW DICEY THIS CAN GET

Content farms suck. We’ve spoken about this before, but a footnote in that conversation grew into a full-blown issue.

Major brands are featuring ads on spam sites with a treasure drive of crappy content. It’s not a good look for Google, which controls the direction of those ads.

Serious ads on crap sites encourage the practice. That’s a problem.

If you’re Toyota, you don’t want the banner ad for your new Camry sitting next to a blog that starts with, “As an AI language model,” that makes your company look like dimwits. 

But Google is struggling to reign in this practice. These AI content mills are sprouting at an uncontrollable pace, and the content is so low-effort that it infuriates readers beyond reproach.

Sometimes the site copies other publications word for word without attribution, which is a clear violation of Google Ad’s policy and copyright infringement. You can find New York Times articles on these sites without so much as a hyperlink.

Speaking of The New York Times, they also wrote a story about AI and ads.

It turns out the future of content is all gloom. AI gobbles up advertising content because it’s rampant and generative video is a strong case and point.

The best text-to-video pieces right now replicate commercials because they are plentiful and consistent. That means AI will make the commercials you see on AI spam sites one day.

It’s all a feedback loop that hurts the bottom line.

You may not read crappy posts by fake writers, but the transformer models that answer your questions do, and that’s bad.

It’s a snake consuming its own tail. The poor output from one AI becomes gospel to the next, and the internet is filling up with low-effort nonsense. 

We can’t train new models on insufficient data. The logistics don’t work. If we stay, the course will hit another AI winter from a lack of quality data to work with on new models.

Advertisements are one of the most plentiful pieces of new, human-made content out there right now, but do we want that to be our future?

If not, we need Google to do better and stop the rampant rise of AI blogs that post thousands of articles every week. It’s hurting the future of AI.

Quick Nuggets

❄️ Snowflake nudges itself closer to Nvidia in a generative AI push

💊 AI-made drug is set to start human trials soon

🏛️ The US House is officially restricting the use of ChatGPT for members

🎲 ChatGPT can tell a story, but it struggles to handle the details of Dungeons & Dragons

📚 You can learn more by leaning into AI for your research

🥵 A new paper argues that language models are fairly weak learners

⚾️ The MLB Draft is coming up, and you can see how AI is influencing the scouting process

🐾 Baidu claims that their chatbot outperforms ChatGPT in some metrics

🩺 Right now, AI helps doctors with paperwork but could do so much more in the medical field in the future

🇺🇸 The 2024 Presidential Race is quickly becoming the first one obsessed with AI

🔥 Fresh Products

• Factiverse - analyze text for mistakes, bias, disputes, & controversial claims (link)

• Slated - AI scheduling platform built for 3+ attendee meetings (link)

• Zing - body sсan to kickstart your fitness journey (link)

• AI Cadences - create AI-driven sales sequences fast (link)

• Unicorn Platform - simple landing page builder (link)

• theGist - unified AI workspace to supercharge humans (link)

• Optic - AI assistant for video calls & work productivity (link)

• Swantide - AI-powered RevOps assistant (link)

Good Content, Classic Cthulhu

There’s something about AI’s ability to turn old-fashion photos into mesmerizing alternate histories that amazes me. Like, you just know that if multiverse theory is real, that this happened somewhere out there, and I am thankful it wasn’t our universe.

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