🤖 AI Deciphers An Ancient Scroll

What's up? You're reading Inclined AI. Wake us up when AI figures out how to fix insomnia, please.

Who am I kidding? I’m already awake and here with headlines:

• AI image detectors are failing at a bad time

• Deciphering an ancient scroll only took us half a year

• Google adds image generation to its AI search tool

• Learn about the No Fakes Act

Be a sponsor of our newsletter.

In the age of 'fake news' and Photoshop scandals, AI image detectors promised to be the knight in shining armor.

But it turns out our digital vigilantes might be experiencing a Spider-Man 3 moment—remember when Peter Parker got a little too carried away and started jazz-dancing down the street? Yeah, it's awkward.

Optic's free AI tool "AI or Not" recently labeled an emotionally charged photo from Israel as AI-generated. 

The problem is that digital forensics expert Hany Farid says the tool is dead wrong.

Farid’s take is like the Sherlock Holmes of pixels: straight lines, accurate shadows, and structural consistencies all point to the image being as real as your aunt's obsession with cat memes.

Now, here's where the digital cobweb tangles.

The wrongful AI tag went viral, setting off a wildfire of accusations that official Israeli accounts were spreading fake news.

Farid warns that this is the chaos that ensues when AI tools, which can operate with up to 90% accuracy, generate false positives or negatives. He says, "Look, it's not CSI; you can't just push a button for the truth."

The takeaway?

AI's current role in image detection is like asking a toddler to solve a Rubik’s Cube—entertaining but unreliable.

As we grapple with questions of AI ethics and real-world impact, remember that our automated tools are far from infallible.

Life-Changing Concepts → gives you an unfair advantage in business using mental models

Master ChatGPT → check out the ChatGPTricks e-book that takes you from zero to hero

Move over, Indiana Jones, there's a new kind of relic hunter in town—and they're wielding neural networks instead of whips!

Think about it: a 2,000-year-old mystery, an unopened scroll turned into charcoal by Mt. Vesuvius, finally gets its big break in the form of ones and zeros.

Nat Friedman, ex-GitHub CEO, and entrepreneur Daniel Gross kicked off the Vesuvius Challenge, and guess who cracked the code?

A 21-year-old SpaceX intern, Luke Farritor, used machine learning to turn ancient scribbles into legible letters. This young padawan snagged a cool $40,000 for his efforts.

Not bad for a night's work, eh?

Federica Nicolardi, a scholar at the University of Naples, thinks this is game-changing.

We might soon be reading ancient grocery lists or even lost philosophical treatises by Aristotle.

Let's face it, folks: AI just made papyrology—the study of ancient writing—the hottest ticket in academia. 

So, Kudos to Farritor for their work!

🎤 AI apps for lip-syncing are starting to shoot up the download charts

👻 Ghostwriter stuns us by showing up, disguised, to the Bloomberg screentime conference

🦾 Roboticists are changing their ideals with more generative AI capabilities coming to light

🤫 Under the radar, OpenAI discretely changes their core values

✨ Google outlines new ways to use generative AI in search

❓ Parents are starting to ask if there’s any way to hide their child’s face from AI crawlers

📜 Take a look at this bipartisan bill called the No Fakes Act, which tackles AI regulation

🔋 EV battery development is quietly getting some AI updates

• Cal - open source AI scheduling assistant (link)

• Akool - swap faces in photos & videos (link)

• AI HomeDesign - re-décor your room w/ 2 clicks (link)

• AICO - create Shorts from a YouTube link (link)

• Infichat - customer support for growing businesses (link)

• Ryan - customized children's fairy tales (link)

• YC Application Optimizer - boost your app w/ AI trained on funded applications (link)

• WorkOutPro - personalized fitness plan in 5 minutes (link)

Rapid Progress

This could be viewed as a visual representation of Moore’s Law but isn’t this incredible? Imagine where we’ll be with AI in just a year or so.

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