Have you heard that author Andy Weir has a new book coming out? Very exciting, we know, and according to a syndicated reading list for Summer 2025, it’s called The Last Algorithm, and it’s a tale of a programmer who discovers a dark and dangerous secret about artificial intelligence. If that seems a little out of sync with his usual space-hacking fare such as The Martian and Project Hail Mary, that’s because the book doesn’t exist, and neither do most of the other books on the list.
The list was published in a 64-page supplement that ran in major US newspapers like the Chicago Sun-Times and the Philadelphia Inquirer. The feature listed fifteen must-read books, only five of which exist, and it’s no surprise that AI is to behind the muck-up. Writer Marco Buscaglia took the blame, saying that he used an LLM to produce the list without checking the results. Nobody else in the editorial chain appears to have reviewed the list either, resulting in the hallucination getting published. Readers are understandably upset about this, but for our part, we’re just bummed that Andy doesn’t have a new book coming out.
In equally exciting but ultimately fake news, we had more than a few stories pop up in our feed about NASA’s recent discovery of urban lights on an exoplanet. AI isn’t to blame for this one, though, at least not directly. Ironically, the rumor started with a TikTok video debunking a claim of city lights on a distant planet. Social media did what social media does, though, sharing only the parts that summarized the false claim and turning a debunking into a bunking. This is why we can’t have nice things.
That wasn’t the only story about distant lights, though, with this report of unexplained signals from two nearby stars. This one is far more believable, coming as it does from retired JPL scientist Richard H. Stanton, who has been using a 30″ telescope to systematically search for optical SETI signals for the past few years. These searches led to seeing two rapid pulses of light from HD 89389, an F-type star located in the constellation Ursa Major. The star rapidly brightened, dimmed, brightened again, then returned to baseline over a fraction of second; the same pattern repeated itself about 4.4 seconds later.
Intrigued, he looked back through his observations and found a similar event from a different star, HD 217014 in Pegasus, four years previously. Interestingly, this G-type star is known to have at least one exoplanet. Stanton made the first observation in 2023, and he’s spent much of the last two years ruling out things like meteor flashes or birds passing through his field of view. More study is needed to figure out what this means, and while it’s clearly not aliens, it’s fun to imagine it could be some kind of technosignature.
And one last space story, this time with the first observation of extra-solar ice. The discovery comes from the James Webb Space Telescope, which caught the telltale signature of ice crystals in a debris ring circling HD 181327, a very young star only 155 light-years away. Water vapor had been detected plenty of times outside our solar system, but not actual ice crystals until now. The ice crystals seem to be coming from collisions between icy bodies in the debris field, an observation that has interesting implications for planetary evolution.
And finally, if like us you’re impressed anytime someone busts out a project with a six-layer PCB design, wait till you get a load of this 124-layer beast. The board comes from OKI Circuit Technologies and is intended for high-bandwidth memory for AI accelerators. The dielectric for each layer is only 125-μm thick, and the board is still only 7.6 mm thick overall. At $4,800 per square meter, it’s not likely we’ll see our friends at JLC PCB offering these anytime soon, but it’s still some pretty cool engineering.
It doesn’t inspire confidence that the SETI article says ‘Project Ozma at the Greenbank Observatory in West Bank, Virginia.’
I’d totally read that book.
Maybe AI is finally coming out of the closet?
I for one welcome our new hallucinating overlords.
$4800 a sq meter is, unless I lost a decimal place, 48 cents per sq cm. A 5x8cm board is $20 assuming you need of sq meter of them. That is 16 cents a layer. Could there be an error?
Anyway, sounds like part of the new Nvidia bandwidth monster. The 120kW fully liquid cooled cabinet of Skynet! The thinner the layer the lower the capacitance and the higher the possible frequency. I bet proto-Skynet did the layout! Like a hermit crab, it has outgrown its shell and needs more.
Yeah, as worded, the HaD is easy to misinterpret…
the linked article mentions the bill of materials cost is $4800 per sq. meter, and a yield rate of 65%. I assume that doesn’t include labor/processing costs either, so presumably, the customer price will be much more
Andy needs to bang out that book fast (using LLM assistance) to cash in on this fantastic opertunity.
Andy is number two on the mostly fake list. If he acts fast enough, he could be talking best seller money.
When it comes to LLM generated content my favourite quote is currently “Why should I waste my time reading something, that someone else was too lazy to bother writing”.
Thinking about this a bit more. The problem is not with using a LLM. The problem is a that a task master allocated the generation of the list to someone who knows nothing about books. That person also knew nothing about the drug induced hallucinations of using any computer LLM. So they idiotically trusted the steaming pile returned as facts instead of half remembered dreams. The task master never allocated resources to review the article. So ultimately you have people using tools that they do not fully understand (And to be fair that is probably true for 99.999 of people using any technology today). So I would see it as a systemic flaw brought about by very poor management. The task should have been allocated to someone who actually reads books. And their article fully distrusted until reviewed by at at least two other people, who are also prolific book readers.
I really miss Bill Hicks: https://m.youtube.com/watch?v=NivJKUfa7Sw
The problem could be a severe shortage of book readers.
Your analysis is spot on.
.
I do … education … and looked into how adults learn and why problems happen. I’m going to muck up the details, but the upshot is something like Toyota studied the problem and basically found that if a system is set up to allow errors to happen, they will. But (for me) more importantly, it is useless to punish the individual making the mistake. In my job it makes sense- unless you are a psychopath and want to come to work to crash a forklift on purpose or something, starting with the assumption that most people don’t come to work to F-up on purpose has been really helpful in my teaching. And more effective. “Root cause analysis” is drilling down and finding the systemic problem.
.
Fixing a systemic issue is much, much harder than just yelling at someone who messed up though, and unfortunately yelling is much, much more common.
.
Finally, as a burnt out teacher, putting in that effort is super hard but I keep doing it and ironically it keeps me actually teaching despite that, because the results are so good.
My husband was a forester turned safety guy who had to write “root cause analysis” documents all the time. The root cause of many of the issues was management. Management hallucinating how fast work could be done, and screaming at crews to work faster because they underestimated the job. Management hallucinating that you could send out crews and work them 14 hour days 6 days a week, and only 10 on Sunday, and there wouldn’t be incidents. Management thinking that training was a magic wand, and if they (who signed the paychecks) yelled at a crew to work faster and the guys cut corners to meet the deadline, that it was the safety departments fault (who did not sign the paychecks).
I have know several bosses I would be happy to replace with a hallucinating AI – I wouldn’t have to pretend to like them, and could swear at the AI without getting fired.
But of course you can’t put “management” in as a root cause.
And kudos to you for continuing to teach. FOr me it would be hard to not yell at today’s class for doing the same stupid thing as last year or last decade’s class…
That sounds fantastically cheap. So a 10 cm square board would only cost $48. That price looks too good to be true.
I suppose an AI board is way bigger than 10 cm sq and also in 10 cm sq you do not have enough space for enough parts to use most of the 124 layers. But I might be wrong.
Maybe it is the bulk manufacturing price excluding mask cost.
Currently JLCPCB offers 32 layers at $58 per 10x10cm.. after you pay the thousand dollar setup fee.
“and it’s no surprise that AI is to behind the muck-up” ?
Not any more.
Rapid flashing star as technosignature? That sounds like bad news
Why anyone can trust AI after people follow Google maps and drive into lakes already is beyond me.
I watched a reasonably intelligent youtube engine builder let AI choose 6 pairs of matching injectors from a list if 20 that had been flow tested, wirh predictably poor results. A variance of three. It took me less than a minute to make 7 pairs with a variance of 1.
I think we need to give the AI better prompts that take into account their deficiencies. Or we need to develop a “common sense” intelligence that fact checks itself and double checks for hallucinations.
Of course if you are intelligent enough to circumvent AI pitfalls, wouldn’t you just perfom the task yourself?
I have met at least two real life people of dubious education/intelligence who basically worship AI, does not bode well for humanity.
I think AI is a very attractive way to deal with tasks for people whose skillset is well below average, because AI seems, to them, to be better than what they’d do.
To anyone who is even slightly above average, AI is probably going to seem easy but of poor quality, and to someone good at a task, AI is going to seem like more work than doing the work.
Debug is usually harder than writing good code in the first place. But if you can’t make something good in the first place, you won’t ever get to the debug step and even if you do, it’s going to go horribly wrong. AI gets you to the horribly wrong step MUCH FASTER, so at least it saves you some time and effort.
I strongly feel that the sentence in the article should read: “Writer” Marco Buscaglia. If it’s written by AI you’re not a writer, you’re a director.
I’d rather have a new book by Andrew Wiles.
Marco Bruscaglio was fired because of the AI listicle, incidentally his LinkedIn says:
“An innovative, experienced and reliable content creator with a background in both journalism and creative writing”.
I guess “reliable content creator” doesn’t technically make any promises about veracity or integrity.
The dielectric layers are 25µm, not 125µm. 125µm would make the pcb at least 15.5 mm thick.
csw