Posts by Simon Harris 2773 publicly visible posts • joined 1 Mar 2007
I remember reading a novel quite a while ago where the process of government had become so automated and efficient that things were happening too fast and a government department existed solely to mess things up a bit and slow things down to a human speed again.
I have a feeling it might have been one of the Niven/Pournelle collaborations, but without going through my book collection I can’t be sure. Can anyone remember the book?
The difference between the LLM that Microsoft is using and a traditional search engine is that a search engine will find pages and you decide how relevant each of the results is to your query. The LLM obfuscates that somewhat by boiling it down with a plausible description - unless it provides a valid list of references (and ChatGPT has been known to make them up), what are the sources it has used to provide that information, and has its prediction method validly come up with the answer?
An example - in a 6502 forum, one member asked ChatGPT what the best method was for breadboarding a 1MHz clock oscillator for a 6502.
ChatGPT suggested using a 555 - now, that’s a classic timer that has been used for decades as an astable oscillator, but 1MHz is pushing it to the limit (possibly past its limit). So ChatGPT had linked the concept of oscillator to the 555, which in many cases would be reasonable.
It then went on to suggest passive component values, and what to connect to each pin of the 555. This uncovered two more problems - the components suggested gave a frequency nowhere near 1MHz, and while some pin connections were correct, others were completely wrong - it’s as if it had learned a pattern for how to describe component values, presumably from a range of online examples, and either the training data was wrong, or it just didn’t have strong network connections to associate the correct components with the desired frequency, and as for the pin descriptions, presumably it learned how to describe connections in general but the prediction mechanism couldn’t make any sense out of the specifics. However, the whole thing was framed in a plausibly written set of instructions, and if followed (rather than looking in a readily available data sheet, easily found with any search engine, that has the actual circuit), you’d be scratching your head trying to work out why it didn’t work.
If you didn’t know it was bollocks you would probably think it was a reasonable answer. When you know about a subject and you see the answer is bollocks, you worry about how poor the answers might be on subjects you don’t know much about.
One problem I think will emerge this year is that we don’t seem to have a consistent approach to how LLM affects education, and how educational establishments should respond to it, with some banning it outright, some embracing it and some somewhere inbetween.
My own considers ChatGPT generated essays (along with those bought from essay farms) a form of plagiarism - and as someone who will be marking project dissertations soon, we need some way to determine what is actually the students’ own work (apparently TurnItIn is working on a ChatGPT detector).
But at the wider lever, we know that among some of the useful output from such systems, there is also some right bollocks. When LLM gives the impression of doing the thinking for us, it can present those bollocks to look as reasonable as the correct answer. We already have problems with people believing nonsense they read online. How do we educate people not to further subcontract their critical faculties to a machine?
I think there’s a bit of a difference.
A boffin, to me, is someone who actually does science.
A geek or nerd may do science, but it might also refer to someone who just has an intense interest in it, or some other subject, not necessarily scientific, for example you could be a language nerd or a history geek. But I’ve never heard boffin used outside of science.
I had an orange plasma Elonex 386sx ‘laptop’ back then (about 1990) - I remember the screen did get pretty hot.
It took so much power they didn’t even bother with batteries so needed plugging in wherever you went, and being somewhat weighty was more of a luggable than a laptop - I think it cost me a little over £1700 - that was the same price as the 286 machine I’d bought at the beginning of 1988 with an EGA monitor.
How did I come by that snippet?
I’m old enough to have watched Blake’s 7 when it first came out and I had a school friend who had a System 1, so it was instantly recognisable.
Incidentally, there is a picture of it here
http://www.starringthecomputer.com/computer.html?c=209
If you really want to scale it, the Acorn System boards were all standard 100x160mm Eurocards
If Chat GPT4 is as smart and can do all the things the ads that keep popping up on my social media feeds claim, why hasn’t it written Chat GPT5 yet?
Maybe I’m just a cynical old git, but I can’t help thinking that all this jumping on the ChatGPT bandwagon reminds me of the dot-COM bubble of the 90s.
Used to do mine like that when I was doing my final year BSc project - a mixture of black tape and 2x Alfac transfers.
Designed with coloured pens on 0.2” graph paper, then overlaid with drafting film and transfers/tape for each layer, and then sent off to be reduced to create the final photo etching masks.
One of my first work projects after my degree was writing a high resolution printer driver for our new PCB CAD system, and I never used the transfers again.
Apple seems to even make itself incompatible with itself!
I have an Apple TV subscription (courtesy of a free gift from my credit card company), an old Apple TV box and a couple of other devices with the Apple TV app. For no apparent reason this week the Apple TV box was the only thing that refused to connect to the Apple TV streaming service even though it could connect to everything else. Weird.
In 1975/76 Wireless World produced a set of articles to build your own Teletext decoder, reproduced here:
https://www.blunham.com/Radar/Teletext/PDFs/WirelessWorldDecoder.pdf
All digital logic without a microprocessor in sight, and you can see from first principles how to decode the signal.
Back in the days before schematic capture and PCB CAD software was affordable, I used to lay out PCBs by hand on 0.2” graph paper - we used to make the originals of the photo-masks 2x size.
Drawn in with 2 colours of felt tip and then, for each side, drafting film taped over the graph paper and the tracks and pads put on with Alfac rub-down transfers.
Once upon a time I had some paper drawings of knee replacement components I needed to turn into 3D models.
Dimensions were in mm, but every one included very odd fractions of a mm. Very odd that is until I converted them to imperial and they all turned out to be nice round multiples of thousands of an inch.
In the old days there were a variety of round pin plugs and sockets of various physical sizes, so you couldn’t plug the wrong class of appliance into the wrong circuit, with 2A, 5A, 15A and 30A being standard. Lights would be most likely on 2A and 5A.
UK theatre lighting still uses the 15A socket as standard.
I remember back in the 70s my grandma still had extension cables with a lighting bayonet connector at one end and a round pin mains socket at the other.
Considering she was still using round pin plugs and sockets then, I was surprised to see that BS1363 goes all the way back to 1947.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
