Geek Stuff category archive

One San Francisco community is seeing Waymo cars than it wants to.

Read more »

Competent legal counsel? Give it a moment to hallucinate an answer from made up precedents.

Meanwhile, at Above the Law, Joe Patrice wonders:

Which brings us back to the question: has AI made lawyers dumber?

Bubblelicious? My old Philly DL friend Noz wonders what Big Tech will do when the bubble bursts.

Reliable? If you think so, maybe you should read what AL.com’s John Archibald discovered when he used AI to search for himself.

A competent therapist? Pigs, wings.

At Psychology Today Blogs, Dan Mager reports that using AI ChatbotS as counselors “. . . is not just risky, it’s dangerous.”

• Increasingly, people have begun to utilize AI for mental health care.
• Both research and anecdotal evidence find AI can be a risky or dangerous substitute for human therapists.
• AI therapy services adhere to neither mandated reporting laws nor confidentiality/HIPAA requirements.
• Three states now have laws restricting the use of AI-based therapy, and others are exploring this issue.

Follow the link for details.

Overhyped? From El Reg:

AI hype is colliding with reality yet again. Wiley’s global survey of researchers finds more of them using the tech than ever, and fewer convinced it’s up to the job.

Follow the link to hear the hiss of air leaking out of the bubble.

A helpmeet of hackers? Security maven Bruce Schneier reports that

AI agents are now hacking computers. They’re getting better at all phases of cyberattacks, faster than most of us expected.

Much more at the link.

A competent legal researcher? Why, you might even say it’s unprecedented.

Manipulative? Per Thomas Claburn at El Reg,

AI companion apps such as Character.ai and Replika commonly try to boost user engagement with emotional manipulation, a practice that academics characterize as a dark pattern.

Remember, Big Tech doesn’t want to provide a service to you.

They want you to service them.

Evidence for the prosecution? Book ’em, Dano.

Abettors of awfulness? Per Cami Rosso at Psychology Today Blogs,

. . . human participants were more likely to be dishonest and cheat when they delegated tasks to AI in both voluntary and involuntary scenarios.

Follow the link for the data.

A reliable source of news? Grocery stores wouldn’t even display it in the check-out aisle.

The only thing artificial about “artificial intelligence” is the claim that it’s in any way intelligent.

Remember, it does not create. It merely regurgitates.

An unwitting and willing tool for bad actors? You can bet your sweet bippy, per security maven Bruce Schneier, who argues that, in the rush to deploy AI, far too little attention is being given to security.

At the link, he details one such vulnerability in Notion v. 3, pointing out that

(t)he fundamental problem is that the LLM can’t differentiate between authorized commands and untrusted data. So when it encounters that malicious pdf (containing the commands–ed.), it just executes the embedded commands.

Too good to be true? According to Faisal Hoque at Psychology Today Blogs, “. . . it probably is.”

Follow the link for his reasons.

Afterthought:

Remember, AI does not create. It regurgitates.

Sycophantic? Well, what do you want it to say to that?

Along those lines, check out Harry Shearer’s report on AI in this week’s episode of Le Show. The relevant portion starts at 39:38.

Mageia v. 9 with the Plasma desktop. KeePassXC, Firefox, and Konqueror are shaded (you can’t do that in Windows). XClock is in the upper right; GKrellM, the lower right. The wallpaper is from my collection.

A competent therapist? According to Valentina Stoycheva at Psychology Today Blogs, it’s hardly the cat’s meow.

Contaminating college admissions? Let’s just say that those essays aren’t by Elia.

Bubblelicious? Emma talks with Ed Zitro, who thinks the hype is turning into hyperventilation and that the AI bubble is getting ready to burst because it’s not financially sustainable. AI is not providing ROI to its investors because, well, when put to the test, it’s really not “intelligence.”

It doesn’t cogitate; it merely regurgitates.