And an additional response, because I didn’t fully answer your question. LLMs don’t reason. They traverse a data structure based on weightings relative to the occurrence frequency in their training content. Loosely speaking, it’s a graph (https://en.wikipedia.org/wiki/Graph_(abstract_data_type)). It appears like reasoning because the LLM is iterating over material that has been previously reasoned out. An LLM can’t reason through a problem that it hasn’t previously seen unlike, say, a squirrel.

By the same logic, raytracing is ancient tech that should be abandoned.

Nice straw man argument you have there.

I’ll restate, since my point didn’t seem to come across. All of the “AI” garbage that is getting jammed into everything is merely scaled up from what has been before. Scaling up is not advancement. A possible analogy would be automobiles in the late 60s and 90s: Just put in more cubic inches and bigger chassis! More power from more displacement does not mean more advanced. Continuing that analogy, 2.0L engines cranking out 400ft-lb and 500HP while delivering 28MPG average is advanced engineering. Right now, the software and hardware running LLMs are just MOAR cubic inches. We haven’t come up with more advanced data structures.

These types of solutions can have a place and can produce something adjacent to the desired results. We make great use of expert systems constantly within narrow domains. Camera autofocus systems leap to mind. When “fuzzy logic” autofocus was introduced, it was a boon to photography. Another example of narrow-ish domain ML software is medical decision support software, which I developed in a previous job in the early 2000s. There was nothing advanced about most of it; the data structures used were developed in the 50s by a medical doctor from Columbia University (Larry Weed: https://en.wikipedia.org/wiki/Lawrence_Weed). The advanced part was the computer language he also developed for quantifying medical knowledge. Any computer with enough storage, RAM, and the hardware ability to quickly traverse the data structures can be made to appear advanced when fed with enough collated data, i.e. turning data into information.

Since I never had the chance to try it out myself, how was your neural network and LLMs reasoning back in the day? Imo that’s the most impressive part, not that it can write.

It was slick for the time. It obviously wasn’t an LLM per se, but both were a form of LM. The OCR and auto-suggest for DOS were pretty shit-hot for x386. The two together inspried one of my huge projects in engineering school: a whole-book scanner* that removed page curl and gutter shadow, and then generated a text-under-image PDF. By training the software on a large body of varied physical books and retentively combing over the OCR output and retraining, the results approached what one would see in the modern suite that now comes with your scanner. I only achieved my results because I had unfettered use of a quad Xeon beast in the college library where I worked. That software drove the early digitization processes for this (which I also built): http://digitallib.oit.edu/digital/collection/kwl/search

*in contrast to most book scanning at the time, which required the book to be cut apart and the pages fed into an automatically fed scanner; lots of books couldn’t be damaged like that.

Edit: a word

No, no they’re not. These are just repackaged and scaled-up neural nets. Anyone remember those? The concept and good chunks of the math are over 200 years old. Hell, there was two-layer neural net software in the early 90s that ran on my x386. Specifically, Neural Network PC Tools by Russell Eberhart. The DIY implementation of OCR in that book is a great example of roll-your-own neural net. What we have today, much like most modern technology, is just lots MORE of the same. Back in the DOS days, there was even an ML application that would offer contextual suggestions for mistyped command line entries.

Typical of Silicon Valley, they are trying to rent out old garbage and use it to replace workers and creatives.

And Cascadia too, please.

our highly trained Canadian Geese

This explains so damned much of their behavior. I for one look forward to these operatives helping us out with our myriad domestic issues.

I’m not a hunter. But I do understand a lot about environmental conservation and the need for balance. We have eliminated enough of the animals that predate on deer such that some other means, ie hunters, are required to control deer populations. The other option is mass kills, which strike me as wasteful on so many levels.

When I lived in Vermont, there was a conservation movement to attract younger people to deer hunting because natural controls just aren’t there anymore. Where I live now, a distemper outbreak decimated the coyotes, and the deer are out of control. The coyotes are finally bouncing back, but it’s going to take a while. In my small city, the deer are so rampant, it’s common to see dozens on a short bike ride through town. Their food supply is depleted enough such that most deer here appear unhealthy and undernourished. The exploded deer population have follow-on effects: increased expense for deer control measures, collisions (one almost slammed into me on my bike two days ago; not the first time), destruction of plantings to control erosion, and spreading ticks.

I would like to see prospering wild animal populations, rather than this mess we made.

the DEA visited them last year and performed “accountability audits” that uncovered violations of the federal Controlled Substances Act, namely through inadequate record keeping, according to records obtained by The Baltimore Banner.

At the scale of prisons, these pharmacies are called institutional pharmacies. The size, operation, automation, and throughput of institutional pharmacies is mind-blowing. For example, the biggest Costco pharmacies might process 300 scrips a day; institutional pharmacies generally handle 15000 to 30000 per day, with some being even larger.

The “inadequate record keeping” part is just idiocy. There exists automation and auditing software for this. I know because I wrote the last-mile portion of a suite that manages end-to-end compliance automation for institutional pharmacies. A single failed audit generally costs more than most of the auditing and compliance suites licensing fees. And even in small pharmacies, there’s usually more than one failed C-2 audit when it happens. And let’s be clear; these audits are always for C-2 drugs (opioids and stimulants).

keeping a product listed that they know is not safe.

Amazon wouldn’t do THAT, would they?

Oh right, they would. https://youtu.be/B90_SNNbcoU And not only would they continue to sell the item, but suppress reviews pointing out the issues.

Anecdotally, six years ago I purchased Ancor marine wiring crimps and 314 stainless steel bolts through Amazon. The crimps were counterfeit garbage and the stainless steel rusted and galled in about two weeks of saltwater exposure. Amazon’s response was basically “contact the manufacturer for warranty.” A quick glance at Amazon listings and it’s clear things have gone further downhill since.

So I regard Amazon doubling down on supply chain fuckery as a net win. I will never shop there again after that hardware BS. And more people will come to the same conclusion that Amazon is quickly becoming the Dollar General of online sales. Add on their shitty treatment of sellers, and good manufacturers go elsewhere, further accelerating the decline.

Which is why they’re not people.

But the C-suite and board are almost like humans. And that’s even better for… things.

The medical field would be categorically fuct. Just the loss of sterile packaging would have serious consequences. Minimally invasive surgeries, joint replacements, bandages that don’t adhere to wounds, stents…

Then let’s consider cordage. Mountain climbing, arborists, rescue teams, sailboats (the most efficient way to cross oceans), ships, construction… the loss of just Dyneema/UHMWPE, which is a relatively new entrant to the cordage field would have seriously negative impacts.

There is a lot of energy bound up in those long molecules, and there are no unexploited niches in balanced ecosystems. There are already bacteria that can consume certain polymers under narrow conditions. Humanity is gonna be so screwed for a long time if bacteria can slip those narrow parameters.

Well, clearly YOU didn’t learn your lesson! :D

They were acquired by Opta Group in 2023. Since then, the quality has declined while prices increased. And around the time of their acquisition, they started doing some shady stuff when claiming USB-IF compliance. The cables were blatantly not USB-IF compliant.

Another example: I personally love my Anker GaN Prime power bricks and 737. Unfortunately, among my friends and peers, I am the exception. The Prime chargers are known for incorrectly reading cable eMarkers and then failing to deliver the correct power. This has so far been an issue for me twice, but was able to be worked around.

This is absolutely by design. The corporate raider playbook is well-read. See: Sears, Fluke, DeWalt, Boeing, HP, Intel, Anker, any company purchased by Vista (RIP Smartsheet, we barely knew ye), and so on. Find a brand with an excellent reputation, gut it, strip mine that goodwill, abandon the husk on a golden parachute, and make sure to not be the one holding the bag.

“It’s what … I do?”

The US really needs a replacement, for example a national digital ID based on PKI… you revoke their access to the account, and everything else that uses the account can keep working

There is already an open standard growing around exactly this concept, Web5 Distributed IDs (DID): https://dev.to/tbdevs/what-is-web5-233o

Disclosure: I worked on the implementation for an Open Banking company (does that need to be disclosed? <shrug> I’m including it lest someone think I’m a shill)

Not to be confused with transcoding, which is important to viewing media.

Got a source on that? According to Backblaze, Seagate seems to be doing okay (Backblaze Drive Stats for Q1 2024 https://www.backblaze.com/blog/backblaze-drive-stats-for-q1-2024/), especially given how many models are in operation.

Heat-Assisted Magnetic Recording. It uses a laser to heat the drive platter, allowing for higher areal density and increased capacity.

I am ignorant on the CMR/SMR differences in performance

My read of it was “the C-suite hates when the engineers actually know how shit works, and the leadership must kowtow to the people doing the actual work.” YMMV or the commentor may have meant something completely different.

This is a serious bummer all around. But wow, does that article suck on its lack of detail. But I guess actually digging into the facts wouldn’t make for clickable headlines. “Oooooh, DRONES!”

1. There are Temporary Flight Restrictions (TFR) in place for that region.

2. Was the drone part of monitoring/firefighting efforts? If it was, that is a terrible error on the part of the sUAS operator and observer. Then again, smoke and fire, which would make for a less interesting story. “Drone participating in firefighting hits plane.” Editor: Boooring! Let’s make it vague so we can cash in on some drone fears.

3. Lots of drones won’t even fly in a TFR zone. More professional drones will warn the pilot AND provide a warning about planes in proximity.

4. All sUAS 250 grams and larger are required to have RemoteID. Plenty of drones won’t even fly unless the RemoteID is functioning fully. And if it shits the bed during flight, lots of drones will just automatically land. Again, except for more professional models or for small cheapies. So one of two things are true: the FAA knows exactly who the responsible party is, or the operator is an utter douchecanoe