Friday Squid Blogging: Catching Humboldt Squid - Source: www.schneier.com

Comments

Anonymous • August 29, 2025 7:33 PM

https://www.yahoo.com/news/articles/identical-twin-convicted-rape-cold-175405227.html

=While submitting additional DNA from the scene, the detectives were faced with an unusual dilemma — the evidence was linked to identical twin brothers John Aurthur Marubbio and Russell Anthony Marubbio.

Since they were twins, investigators were unable to differentiate between the two and identify which one was the suspect.

While identical twins start with the same DNA, rare mutations can arise after the twins split in early development, creating subtle genetic differences.

Parabon was able to analyze the samples and conclusively identify Russell Marubbio as the twin whose DNA was at the crime scene.=

Clive Robinson • August 29, 2025 7:53 PM

@ ALL,

Are tech giants getting nervous? They should be

Is the sub title of an opinion piece in The Register.

https://www.theregister.com/2025/08/25/overinflated_ai_balloon/

Which gives an “AI Bubble” report card of,

“Mediocre at best, on the way down, and never going to get an A grade now.”…

Hence the main title of,

“The air is hissing out of the overinflated AI balloon”

Which to some of us is not exactly news.

It quotes a report that marks Current LLM systems down as being at,

“The not particularly bright or trustworthy intern”

“bottom end” level/Grade…

Pointing out,

“There’s a whole shadow world of people using AI at work. They’re just not using them “for” serious work. Instead, outside of IT’s purview, they use ChatGPT and the like “for simple work, 70 percent prefer AI for drafting emails, 65 percent for basic analysis. But for anything complex or long-term, humans dominate by 9-to-1 margins.”

Before the article author asks the all to obvious

“Why?”

Before saying,

“Because a chatbot “forgets context, doesn’t learn, and can’t evolve.” In other words, they’re not good enough for mid-grade or higher work. Think of them as a not particularly bright or trustworthy intern.”

Then stating a painful truth that even the dumbest of investors is going to realise before long,

“That may be good enough for $20 a month, but – spoiler alert – AI costs will have risen by ten times or more by next year.”

Then asking the “ROI or Not?” Question,

Will bottom-end AI be worth that to you? Your company?”

The answer being very likely “Definitely NOT!”.

The article ends as some have suspected, by talking about the Tech Stocks crash back in the 1990’s that wiped out many small investments, and goes on to end,

“Most companies, though, have found that AI’s golden promises are proving to be fool’s gold.

I suspect that soon, people who’ve put their financial faith in AI stocks will be feeling foolish, too.”

I can’t argue with that prognosis, because it’s really just a question of time as history repeatedly shows…

ResearcherZero • August 29, 2025 10:30 PM

@Clive

RE: AI’s golden promises are proving to be fool’s gold.

Part of any successful insider trading scheme is to squeeze as much fat out of the pig as possible and ensure you are in a position where you cannot be prosecuted. 😉

Some might lose their jobs and homes when the entire house of cards comes down, but they will be too busy dealing with day to day survival and too poor to mount any response.

Everyone else will be too busy blaming the bankers to figure out what exactly took place.

lurker • August 30, 2025 1:08 AM

@Clive Robinson, ALL

Has Taco Bell seen the light, or just cutting their losses? For light amusement, search terms:
taco bell drive-thru AI

KC • August 30, 2025 1:14 AM

@Clive

I’m changing my handle to KC from Anonymous to lessen any confusion here.

To add to your commentary on AI, it’s being
observed that newer models are getting more expensive, not less.

Although the cost per token is going down, the number of tokens used for so-called reasoning is going up.

Many newer AI models double-check their answers, go out to the web for more intel, and so on. But the improved quality comes at a cost.

To avoid price shocks, many consumers could be nudged towards “dumber” AIs. But other businesses, such as those that use AI to write resource-intensive code, could see a decline in margins or worse.

I use the free versions. Maybe it’s best I don’t know what I’m missing.

I sent a family member an article about an AI discovering new physics. My hunch is that those scientists had one of the less dumber models.

https://phys.org/news/2025-08-ai-reveals-unexpected-physics-dusty.html

Subscribe to comments on this entry