Use AI to talk to pets?

This week in AI

1. OpenAI strikes deal with Vogue owner: things changed from lawsuit to deal real fast
2. Unstoppable rise of Chubby: this AI-gen cat can break the internet?
3. GPT-4o turns a PDF earnings report into an interactive dashboard in 30 seconds: productivity goes 100x

How do AI checkers keep things original?

Back to the topic….

If you're trying to pass off AI-generated text as your own, you're in for a surprise.

Teachers, editors, and publishers are using smarter content detectors every day. It's time to put in the real work.

The Technologies Used by AI Checkers These AI checkers use different smart techniques to catch even the trickiest attempts to deceive:

1. Text Analysis:

? They break down the text into individual words and phrases, looking for patterns—both common and uncommon—that indicate whether it was written by a human or generated by a machine.

? Things like unusual word choices, awkward phrasing, and a lack of personality are red flags. For GPT, words like “delve” and “realm” are some of those.

2. Contextual Awareness:

? They examine the overall context and semantics, trying to understand the meaning behind the words rather than just the surface-level structure.

3. Cross-Referencing:

? The real power of these tools lies in their ability to cross-reference submitted text against massive databases of online content.

? They can rapidly search the internet for matching passages, close paraphrases, or similar wording. The larger the checker's database, the better it can detect text that's been copied or artificially created.

Now, I get it—the temptation to take shortcuts is strong, especially when faced with tight deadlines or challenging assignments. But the consequences of getting caught simply aren't worth it.

These AI detectors are only going to get smarter and more sophisticated over time.

Read more about it

Chatting with Chimps? AI Might Make It Happen

Way back in the 1980s, elephant researcher Joyce Poole noticed something intriguing about how elephants communicated.

When one elephant would call out to her family, only certain individuals would respond, while others would ignore the call. This suggested that elephants might have a way of directing their calls to specific herd members.

Decades later, Poole partnered with AI researcher Mickey Pardo to investigate this hypothesis. They recorded elephant calls in the field, carefully documenting who was making each call and who it was addressed to.

Read the hypothesis

This is just one example of how machine learning is helping researchers decode the complexities of animal communication that humans can't easily detect on their own.

And now, AI researchers want to take this a step further, aiming to develop large language models tailored for interspecies communication.

Translating Across Species

? The Earth Species Project is exploring whether animal communication systems have an underlying "shape" that could enable translation without labeled data.

? The goal is a kind of interspecies communication based on aligning the representations of animal sounds, images, and other sensory information.

? Challenges remain in validating these models and managing expectations around true shared understanding across species.

For more insights,

Watch this video

AI tools of the week: My top picks

There are a lot of productivity tools; here are my top picks for your workplace:

Superhuman

Gumloop

Sune