Let's talk tech Thursday #15
Welcome to another edition of Let's talk tech Thursday, the newsletter that understands that days of the week are a social construct.
This week we:
- Examine what happens when AI wins legal battles against creatives
- Think about the ramifications of Hawk-eye replacing line callers at Wimbledon
- Pay another visit to Manchester's retail scene, where Iceland are rolling out facial recognition technology.
Also: autonomous busses, and digitally scanned actors.
Let's dig in...
Top Stories
Anthropic wins key AI copyright case, but remains on the hook for using pirated books
Summary
A US judge ruled that AI firm Anthropic can use books to train its software without breaking copyright law because the use is "quintessentially transformative." However, Anthropic must face a trial over keeping pirated copies of books in a central library.
So what?
In one of the first cases of it's kind, the American legal system agreed that AI learning from existing works is not a breach of copyright. Instead, this constitues fair use, in the pursuit of turning a "hard corner" and create something different.
It's an interesting case, because it seems that where the authors' arguments fell down - at least according to Judge Alsup - was that they weren't trying to argue that the Large Language Models were creating knockoffs of their work. Had they done so, speculated Alsup, this would have been a different story.
Presumably the reason they didn't try to go down that road is that it's harder to prove (though I'm not a lawyer, so that's speculation on my part). But if they could have, it strikes me that would have been an easier case to win. We know that Large Language Models are designed to reshape vast quantities of information into different, but understandable, shapes. It would therefore be relatively easy to take an LLM that had ingested (e.g.) all of Andrea Bartz's work and ask it to produce a novel based on her style of writing - perhaps even using some of her characters.
But perhaps this is where we get into tricky grounds. I can, almost verbatim, recite any and all of Douglas Adams' works. Other than being the life and soul of dinner parties, I don't do anything with that. But if I wrote a book pretending to be Adams I would - rightly - be sued by his estate. Is this the same thing with AI? Just because you could get Claude to recreate Bartz's work, does the fact that you don't mean it's fine that all of her works just sit in the LLM? And even if someone did, presumably it would be the individual entering the prompt that was liable, not Anthropic? Adams' estate wouldn't sue Waterstones if I started plagarising his work, after all.
Crucially, case was discussing works obtained by Anthropic lawfully. It's worth noting that an estimated 7 million books sit in the LLM that have been pirated. These, Anthropic will still have to answer for as part of separate court proceedings.
Whatever your stance on this, we've now entered a state of precedence in the AI vs Creatives lawsuit. This might have been one of the first, but it certainly won't be the last, and it will be interesting to see how this case shapes all the others.
Will there be line judges at Wimbledon 2025
Summary
Wimbledon will not have line judges in 2025, using electronic line-calling technology instead. Supporters say the system will be faster and more accurate, while critics will miss the tradition and human touch. Chair umpires will remain to manage the game and support players.
So what?
Something that really caught my eye with this story was a quote about half way down.
"Why would you go to call the lines at Finchley Tennis Club under-12s if you haven't got the carrot of 'maybe one day I can get to call lines at Wimbledon'?"
Pauline Eyre, who called lines at Wimbledon for 16 years, is articulating almost the same point I'm seeing at technology companies all over the country. If we replace today's "junior staff" with technology, who is going to be tomorrow's senior staff? Except she's pointing out the reverse effect. If you can't be promoted to a role, why would you start at the bottom of that ladder?
It's not a side of the coin I'd really considered, but since reading that article, and Eyre's quote, I've been wondering about some of the other impacts AI adoption is having on the job market. Is there an AI parallel to the under-12s line callers? Are there sectors young people aren't going to college or university for because the entry level job market doesn't - or soon won't - exist? Or is there a more lateral comparison? If big profitable companies are using AI for certain things, will the smaller feeder companies stop functioning?
The 2023 SAG-AFTRA strike in Hollywood might shed some light on this. A big part of the protest was around the use of AI to replace background actors. Background and small parts are how many actors start their careers, and with fewer entry-level jobs available, you have to wonder at the level of representation that's possible in film and TV.
What else is going on?
Frozen foods supermarket chain deploys facial recognition tech
One of our top stories last week was Manchester's Danielle Horan who was incorrectly banned from a number of Home Bargains stores as a result of a facial recognition mix up. This week, frozen food retailer Iceland are looking to implement the same Facewatch technology - also starting with a Manchester-based trial.
Iceland claims it will not store data of innocent shoppers, but it's worth noting that - ignoring the fact that Ms Horan was innocent and had her data stored - the system needs to store all faces captured, at least temporarily, in order for the processing to work.
Sticking with stores of facial scanning and consent (or lack thereof), nearly 1500 actors have signed an open letter calling for increased protections against the use of AI in film and television. This comes after thousands of UK actors have been "digitally scanned" without informed consent. Details are thin on the ground as to what the digital scanning was for, but it's not a huge leap to imagine the kinds of things actors might be worried about - from their data being used to feed AI models, or perhaps the creation of "actor-less" content.
Self-driving busses open to passengers for free journeys around Cambridge
Self-driving buses are now carrying passengers for free in parts of Cambridge. Driven completely autonomously, but with a "safety driver" on-board in case things go wrong, these busses are currently running a route from Madingley Park & Ride through the Cambridge West Innovation District, with plans to extend if the pilot is successful.
The director of Whippet buses has said that he doesn't see this as posing a threat to bus drivers' jobs. It's hard to see how this could be true, though I suppose if there are routes that are currently too expensive to operate then this might be a cheaper alternative. Companies though, famously, don't often continue to run expensive parts of their business if a proven cheaper alternative exists. It remains to be seen.
A big week for AI in this edition of LT3. And on reflection not a particularly positive one. There is definitely other tech out there though, so next week I'll try and bring you a slightly more varied menu.
What kind of stories do you want to read more of? I'm always keen to hear from you. Drop a reply and let me know!
In the meantime, have a great weekend, and I'll see you next week. Possibly on Thursday.
Will
