HomeIndustriesDAI#27 – Video magic, crazy AI and Chinese deepfakes

DAI#27 – Video magic, crazy AI and Chinese deepfakes

Welcome to our weekly roundup of top AI news.

This week OpenAI magically turned words into videos but lost the plot.

Google has gone big with one other Gemini release.

And giant AI-generated rat testicles have made it right into a research paper.

Let's dive in.

Sora in your loss Google

Our 2024 AI wish list was fulfilled in dramatic fashion when OpenAI introduced us to Sora, its advanced text-to-video model. The developers at OpenAI chuckled when Google announced Lumiere a couple of weeks ago, knowing full well what they’d bring to market with it.

“Oh, your model makes 5-second clips that look pretty realistic? Cute.'

The OpenAI employees with shares in the corporate are in for an enormous payday. OpenAI has accomplished a young offer that can see the corporate's valuation reach $80 billion, 3 times what it was 10 months ago.

There wasn't all excellent news for OpenAI this week. ChatGPT went from lazy to crazy when the chatbot spewed out a series of strange outputs in an odd temporary breakdown. It could be nice to know why this happened, but OpenAI has not commented.

Gary Marcus is a little bit of a crybaby with regards to AI dangers, but given the present madness of ChatGPT, he raises some valid questions.

Google in context

With the announcement of the discharge of Gemini 1.5 Pro, Google continued its pursuit of AI relevance. Are you having a tough time maintaining with different names of Google's AI products? Yes, we aswell.

Gemini 1.5 Pro beats Gemini Ultra in most benchmarks, but the true news is the model's 1 million token context window. This huge context window combined with impressive recall is a game-changer, especially for programmers working on large coding projects.

Much of Google's revenue comes from promoting revenue, which relies heavily on its search tool and display ads. Both functions are threatened by the Arc AI browser. This could possibly be the brand new way we search the web, but there are some risks that have to be addressed.

Could Google be the mysterious company involved in Reddit? Reddit has signed a $60 million-a-year cope with an unnamed company to license access to its user-generated content.

Boys against girls

The AI ​​confirmed what we knew all along; Men come from Mars and ladies from Venus. A Stanford AI model accurately distinguishes female and male brains by analyzing brain scans. It has interesting implications for neuroscience, psychiatry, and fascinating together with your partner.

“I just don’t understand you!”
“Here babe, read this research paper.”


Perhaps these researchers can explain why AI models on OnlyFans earn 10 times greater than the common American.

Biology researchers had a blast with AI by inserting bizarre AI-generated images into the papers they were writing.

The paper was apparently peer-reviewed before being published within the Frontiers Journal, but we're unsure how the enormous AI-generated rat testicles escaped their scrutiny.

A brand new AI tool will make reading these massive research papers much easier. Adobe launches a brand new PDF AI chatbot. AI Assistant can be integrated into its Reader and Acrobat products and can have an enormous productivity boost for anyone who needs to know long documents.

Fight against counterfeits

There has been quite a lot of speak about fighting AI counterfeiting, but the issue is barely getting worse. This week, a large deep fake campaign targeting the US election got here to light. It looks like China is behind it, but the pictures are clunky and folks won’t even pay much attention to them.

Does the reported low engagement mean deep fakes usually are not as dangerous as we predict? Sam explores the surprising impact deep fakes can have on our memory and significant pondering skills.

Big tech corporations are joining forces to crack down on deep fake election campaigns. But will adding security measures to their tools be enough?

Some experts consider the answer to AI security is to bring hardware regulation into the combination. SoftBank's CEO just announced that he’s planning a $100 billion AI chip project. It can be difficult to control who ultimately gets access to those chips as more manufacturers enter the fray.

I imagine a future where a shady guy in a trench coat says, “Psst.” Hey buddy. Want to purchase some NVIDIA GPUs?

In other news…

Here are another clickable AI stories we liked this week:

And that's a wrap.

Has Google's Gemini Pro made you reconsider your ChatGPT Plus subscription, or are you waiting for GPT-5 with Sora built-in? That could be great!

While we're not saying you're on OnlyFans, for those who are, would you hypothetically care if the person was AI-generated? Maybe we'll see human OF models on the picket line soon.

The 12 months 2024 has barely begun and we have already got Sora. Can you imagine what AI technology we can have by the tip of the 12 months? Spirit. Blown.

If you discover a cool Sora video, please send it to us. And tell us if we missed any exciting AI news that we must always have covered.


Please enter your comment!
Please enter your name here

Must Read