Re: I asked ChatGPT for a Dylan Thomas style poem about AI, and it's a weird lament...
Hmmm. The scansion's a bit arratic.
40471 publicly visible posts • joined 16 Jun 2014
Take the work of any poet - or, come to that, any prose - from a few centuries ago and there will be changes of grammar and vocabulary since it was written. The reader may no instantly grasp this. Unless the LLM has been trained on Shakespearean English it's going to come up with something that requires less thinking about for a modern reader.
Just the first 4 words of Shakespeare's Sonnet are enough to highlight some differences:
"Shall..." Oops. Wouldn't a modern writer be more likely to write - and a modern reader expect - "Should"?
"Shall I compare thee..." What? 2nd person singular pronoun? How is that to be understood? Growing up in rural Yorkshire that was a familiar expression from the older generation and the protocol was internalised (it's the same as the use of "tu" etc. in French) but archaic today. Would a modern reader instinctively grasp the intimacy of the phrase? Would and LLM select it and if so, automatically use "Thou" on the next line?
And that's only the easy bit before the mind-bending nature of the comparison that's being made, the thought that went into working it out than that would then be demanded of the reader even given that, for instance, a summer lease might be a more common concept in his time.
An alternative would be to patch a couple of incoming calls together and let the scammers try to scam each other.
I sometimes used that approach with email scammers:
"That sounds most interesting. I'm too busy to deal with it personally at the moment. Could you please liaise with my colleague at $AnotherScammersAddress"
After all, they're in the same line of business so it's only polite to introduce people with common interests.
"However, given life has occurred here"
Were it not for the counter-example you'd have to consider so unlikely as to be impossible.
Consider the number of different systems integrated to compose life. You have the RNA-protein synthesis system which depends on having not only the RNA template but also the amino-acid specific transfer RNAs and the amino-acid/RNA specific activating enzymes.
I'll concede that the ribosome might not be necessary and could have evolved later and the same applies to DNA. You still need some mechanism to replicate all that RNA.
That's a lot of proteins, all of which have t be coded for in RNA. Given that RNA itself can catalyse peptide bond formation it's still a fearsome boot-strapping problem if you're powering it by thermal energy and an even bigger bootstrapping problem to find some naturally-occurring chemical energy source that can link to it. If you want to see it evolve beyond consuming whatever stocks of organic chemicals that non-life processes can provide you also need some form of photosynthesis to evolve.
You also have to have the planet not only become suitable for life at some stage in its development, it's got to stay that way whilst all this unlikely bootstrap happens. Given that photosynthesis is going to involve photolysis of water that's going to require a means of dealing with the release of something as thoroughly nasty as oxygen.
It's the improbability of all those things being strung together successfully that you have to set against the number of planets.
It's a big ask and, I think, reasonable to discount until a second example is found. It's also possible to understand the thinking behind the "life arrived in a comet" style of thinking about lefe on Earth - it's a throwing the problem over the wall solution.
sentient alien life exists in our vast, nearly infinite universe.
FTFY
Sarcasm aside this is a statement for which we have evidence and, if you want to insist, you can remove the word "sentient" but with the word "alien" is in there there is no supporting evidence and something I find very unlikely.
The one that decided we all needed a course on "Empowering a quality culture: strategies for excellence", springs to mind.
I think you may have misunderstood the motivation. The course wouldn't have cost that much and leaving of your own accord they don't have to pay redundancy money.
"Back in 2012, the government created a legal obligation for energy suppliers to make sure they completed the rollout of smart meters by the end of 2019. Subsequently, it pushed back the deadline three times, first to the end of 2020, then 2024, and then 2025. As of February 2023, the government launched a consultation on plans to have smart meters installed in 80 percent of homes and 73 percent of small businesses by the end of 2025."
They should, by now, realise that there's a substantial customer resistance, that they've got about as far as they're getting, that it was a Bad Idea and they might as well quit while they're behind.
I suppose what stops this from dawning on them is that it looks just like so many other HMG projects - behind schedule and over budget - that they think that with time and budget it will get there in the end if they really want it.
"Here in the UK we outsourced everything and their dog"
And the way things are going we're outsourcing more and more food production in favour of solar farms, data centres etc. I trust it will be ministers, heads of quangos and tech CEOs who will be the first to give up eating.
"But if needs be, the "financial services" could be on a parallel terminal from postal services or general sales."
The thing is, there are a LOT of back-end services and not enough space for all the parallel terminals. It has to act as all those parallel terminals. I suppose these days a browser directly accessing all those back-ends would be the way to do it but the services would still have to be written by someone with a bit of programming nous, say enough to know not to commit half-complete transactions.
I think the issue here is publishing something as FOSS - or maybe "jumping on the FOSS bandwagon" would be a better phrase - without thinking it through. Failing to have done so is no excuse for behaving like a sulky kid.
If he'd started off with a more restrictive licence than it wouldn't have been such an issue; a different one, maybe but not one that resulted in writs.
"And it's standard operating theatre practise - you count everything (instruments, swabs etc) before starting and count everything at the end to make sure you've not left anything in the patient."
A week ago after dental treatment. Dentist & assistant counting the instruments. Me, sitting up but looking down: "There's on of the floor".
"found a box-end wrench I'd mislaid and forgotten."
Embarrassed garage owner: A few hundred yards after collecting my car after a service it developed a loud rattle from the front O/S wheel. Returned to garage where the hub-cap was removed along with the socket spanner that had been left on a wheel nut.
"Reputation can be an important thing to researchers and those looking for other research to build from."
Accord reputation on the basis of openness to solve that problem. In fact, natural selection may eventually help. As Universities get more an more strapped for cash and cut down on library subscriptions it will be articles in open on-line journals that get cited more and the less open ones become backwaters to avoid.
Then there's JStor. I did a quick check and they have papers I wrote long ago on there. I can pay to download a copy of my own work. Did they ever ask for permission to take money for them? Of course not. The really annoying thing was that at least one of them was in a free publication at the time.
It might be a case of the users not actually grasping how they'd use it if you just told them about it - and even if the customers had been consulted it would probably have been the management who answered. With hindsight they should have made up a few prototypes and let the end users loose with them. Even then it might not have worked as from TFA only some branches came up with taping the knife to the reader so they might not have learned from that.
"There was a time where the browser space was absolutely the front line of an open, safe, equitable internet," our source said. "Mostly that was about, frankly, preventing Microsoft from taking over the internet."
That worry faded over time, and four years after the 1.0 release of Firefox, Google Chrome debuted.
So now the worry's about Chrome taking over the internet.