Why OpenAI Should Worry, Plus AI Ousts a CEO and Brain Blobs for AI

Greetings, AI aficionados!

As usual, much is going on in the world of AI today. Here is a quick rundown of the headlines:

• Why OpenAI Should Be Worried: The Open-Source Tide

• A Major CEO Loses His Job to AI?

• An AI Investment Surge, with More to Come

• Unintentionally Hilarious: GPT Doing the Gemini Demo, But Real

• Researchers Create "Brain AI"

• Using AI to Raise Talk to the Dead

Enjoy today’s news, and check in on www.intelligist.ai for more!

Joaquin

Why OpenAI Should Be Worried: The Open-Source Tide

Mistral might not beat out ChatGPT in the benchmarks, but it is showing how powerful open source is.

This France-based startup is open-source and just raised over $400 million in funding while being valued at over $2 billion.

Its LLM, Mistral, has shown the power of open-source development. ChatGPT is approximately 5-6x older, and in just a couple of months, Mistral has exploded and become a competitor to everyone's favorite LLM. Is this due to the power of teamwork, or what? 😇

And now, Mistral has released "La plateforme," its API endpoints, which directly challenge OpenAI's. What's more, Mistral's service, already claimed to beat the popular GPT-3.5 in benchmarks, is significantly, fractionally cheaper.

Translation for the non-tech folks: there are now two versions of Mistral. One that can be run locally using your hardware. One that can be accessed from within an app using API calls (Mistral takes care of the processing, hardware, network, etc.). The API allows developers to build portable apps that can work on any device and offer them for sale, paying for the actual Mistral resources used.

The lightning-fast run of Mistral to unicorn status is impressive. But look past the dollars and cents. What we have is an example of what transparent, open-source AI can accomplish. If things continue this way, in a couple of years, we might be talking about ChatGPT the way we talk about MySpace today.

A Major CEO Loses His Job to AI?

This might be the second time a CEO has lost a job to AI, the first being Sam Altman (though that was quickly nixed).

Ross Levinsohn, CEO of the Arena Group, which owns Sports Illustrated magazine, was removed from his position this week.

This comes after last month’s scandal, in which it was revealed that the legendary American sports magazine had been using fake, AI-generated writers to produce its articles. This revelation did not sit well with SI employees or readers, prompting calls for heads to roll.

While the company has not mentioned the recent AI scandal at SI, it is impossible to think that the scandal can be disassociated from Levinsohn’s tenure. In a sense, it is safe to say that, in the case of Sports Illustrated, AI played a role in this CEO shakedown, though not in the way many might have imagined it would.

An AI Investment Surge, with More to Come

For those playing the numbers game, AI investments are getting hotter and hotter. Earlier, we mentioned that French AI company Mistral already has a market valuation of over $2 billion, though they are not publicly traded at the moment.

Meanwhile, reports state that Nvidia's piggy banks are so full that they are using their surplus to not only expand its operations to Vietnam, but invest in AI startups. The California hardware maker has invested in at least 35 AI companies this year alone.

Meanwhile, another area that is hot in investment is medical AI. Recently, AstraZeneca upped its investment in AI-generated drugs, which are already a thing. The company announced a $247 million investment in AI startup Absci.

Unintentionally Hilarious: GPT Doing the Gemini Demo, But Real

Perhaps you haven’t heard about it, but Google caught quite a bit of flak for releasing a “big deal” demo of Gemini, its new AI platform. The demo featured a smooth and beautiful video of a person interacting (supposedly) with Gemini. The catch was that the demo was highly manipulated and, essentially, faked. It exaggerated Gemini's speed, removed any processing delays, and even did a human voiceover of Gemini’s voice.

A coder with a sense of humor Greg Sadetsky, better known as Greg Technology, decided to code an app for GPT-4 to see if it could outdo Gemini. The result is a hilarious (and telling video) entitled ““ and that can be seen here.

The lesson from all this: AI is going to change the world, but it will also unleash a torrent of bulls**t, exaggerations, lies, and scams.

Researchers Create "Brain AI"

Researchers at IU have cooked up "Brainoware," a Frankenstein-esque blend of human brain bits and electronic circuits.

This creation blends brain tissue with electronic circuits and AI technology. It uses brain organoids with neural networks and has succeeded in voice recognition tasks (with 78% accuracy). The researchers claim to be looking for new avenues to advance neuroscience and AI, especially for modeling brain disorders.

But wait-it gets more grotesque! The researchers admit that keeping their little brain blobs alive is a challenge. Neuroscientist Lena Smirnova stated that keeping the organoids alive is a challenge that requires incubators and that bigger organoids will be required for more complex calculations. Cue horror sci-fi visions of a lab with tubes of floating brains suspended in plasma incubators. You heard it here first.

Using AI to Raise Talk to the Dead

And in some even more bizarre news, the New York Times reports that a new startup app aims to allow you to talk to dead people. The new app, HereAfter AI, is trained on content created by a person who is no longer alive. The app then takes that data and uses it to respond to users, in some cases even using the voice of the deceased.

Another startup, StoryFile, produces interactive videos of nonliving individuals and allows users to interact with them. What do you make of this? Is it sci-fi horrible/dystopic? Is it legitimate therapeutic use of AI? Your thoughts, please!

That’s all for today! If you’ve read this far, enjoy your reward: a naturally intelligent brain absolutely bursting with potentially useful (or useless) information! Until next time!

Joaquin / the intelligist.ai