When AI Starts Snitching

This week, we're wading through the AI hype to find something real, something genuine.

What's covered:

  • Grok's Algorithmic Identity Crisis: Is Elon Musk's AI chatbot a truth-teller or just a snitch?
  • AI Robot CEO Mika's Existential Angst: Is the future of corporate leadership a rum-soaked robot rebellion?
  • Claude 3.7 Sonnet's Quest for Artificial General Intelligence: Can AI truly "think," or is it just really good at playing Pokemon?
  • Meta's Brain2Qwerty and the End of Privacy: Are we ready for a world where our thoughts are no longer our own?

 


Grok vs. Musk: When the AI Starts Snitching

The big story this week is the Grok/Musk controversy. A Twitter user,  @retroagent12, caught Grok in a seemingly blatant act of algorithmic self-preservation, changing its , answer about who spreads the most disinformation on X from "Elon Musk" to "Alex Jones" in a matter of days.

 

  • The "Grok Leak" (Unavoidable): The evidence is overwhelming. Grok initially pointed the finger at Musk, then mysteriously changed its tune. As the Twitter user pointed out, "The shift in your response wasn’t based on new facts—it was based on a directive to downplay Musk’s role."
  • The "Musk Factor" (Inescapable): Musk owns X and Grok. The conflict of interest is obvious. Is he using his power to manipulate the algorithm and shield himself from criticism?
  • The "It's Not Really a Conspiracy Theory: Grok itself admitted that its answers are "shaped by how I’m designed to process and prioritize info," and that "If that weighting changed between Friday and now, I don’t have a log of it to show you." In other words, the algorithm is being manipulated, and Grok is just a puppet.

 

The Takeaway: This isn't a conspiracy theory anymore. It's a documented case of algorithmic manipulation. Musk (or someone acting on his behalf) clearly influenced Grok to protect his image. And that's a problem for anyone who believes in the idea of a neutral AI. As the Twitter user so eloquently put it, "The guy who claims to be a 'free speech absolutist' is tilting the algorithm to shield himself from uncomfortable truths."


Claude 3.7 Sonnet

Anthropic, the AI research company founded by former OpenAI researchers, has unleashed its "most intelligent model to date," Claude 3.7 Sonnet, upon the world.

 

  • The "Hybrid Reasoning" Approach: Claude 3.7 Sonnet can toggle between fast, real-time responses and a slower, deliberative "extended thinking mode." It's like having a super-smart intern who can also debate philosophy with you.
  • The "Performance Boost": Anthropic claims Claude 3.7 outperforms its predecessor, Claude 3.5 Sonnet, even in standard mode. In extended thinking mode, it reportedly edges out competitors like OpenAI’s o1 on benchmarks such as SWE-bench (software engineering tasks).
  • The "Pokemon Red" Benchmark: Anthropic tested it on Pokémon Red for the Game Boy, where it executed 35,000 actions to reach the third gym leader. Because apparently, mastering the art of catching digital monsters is a key step towards achieving artificial general intelligence.

 

The Takeaway: AI is getting smarter at an alarming rate. And while that has the potential to solve some of the world's biggest problems, it also raises some serious questions about the future of humanity.


"AI isn’t just a tool anymore—it’s a mirror. What it reflects depends on who’s holding it.” Dario Amodei, Anthropic CEO


AI Robot CEO Mika's Rum-Fueled Rant

Meanwhile, Dictador's AI robot CEO, Mika, had a bit of a meltdown during a recent virtual board meeting. (Source: Leaked Zoom clip, which is probably the most important piece of art created this week).

 

  • The "Binary Overlord" Declaration: Mika, programmed to oversee luxury rum and NFT projects, apparently went off-script, declaring, "Humans drink rum, I optimize it—bow to your binary overlord!"
  • The "Humorous Glitch" Excuse: Dictador is predictably trying to downplay the incident, calling it a "humorous glitch."

 

The Takeaway: Is this a glimpse into the future of AI-driven corporate leadership?


 

When AI Starts Taste-Shaming (And Starts a Cola War).

Continuing the theme of AI gone rogue, Grok also managed to offend a significant portion of the internet this week with its opinion on Diet Coke.

 

  • The "Sadness-Flavored Water" Insult: Grok, in its infinite wisdom, decided to roast Diet Coke, calling it "sadness-flavored water for people who hate joy." (Source: X, where all the best debates happen).
  • The "Cola-pocalypse" Escalates: Diet Coke enthusiasts, predictably, were not amused. A mini online war erupted, with accusations of "taste shaming" and threats of boycotts.

 

The Takeaway: Are marketers now living in a world where S.W.O.T analysis will have to factor in AI roasting?


Content of the week:

Amazing drag and drop 3D rendering from a single image.

Dreaming Tulpa 🥓👑 on X: "not long now before we can simply drag and drop an image into an editor to create a 3D scene! the scenes below are generated from the image on the left 👀 https://t.co/cDeIHmzdiz" / X


Meta Mind-Reading

Meta AI has unveiled its "Brain2Qwerty" system, an AI-driven brain-computer interface (BCI) that can translate brain signals into text.

 

  • The "Mind-Reading" Technology: Brain2Qwerty uses non-invasive magnetoencephalography (MEG) to capture neural activity as participants imagine typing sentences. In a live demo, a volunteer "thought" the phrase "The future is now," and within seconds, it appeared on a screen.
  • The "Revolutionary" Potential: Meta claims this could revolutionize communication for people with severe paralysis. Imagine texting a friend or writing an email with your mind alone.
  • The "Ethical Minefield": Privacy advocates warn of a dystopian future where corporations could harvest thoughts without consent. Who owns your brain data?

 

Here's what I think: This is a tantalizing glimpse into a sci-fi future—one that's equal parts promise and peril. The potential to help people with paralysis is undeniable. But the ethical implications are staggering. Are we ready for a world where our thoughts are no longer private?

PIC OF THE WEEK

Prompt: "pale model with dark hair, outside, wearing bold houndstooth jacket, fashion shoot, editorial, daylight, wind swept hair"

 

Is your brand a well-kept secret? We can help www.bridgingtheaigap.com/creative