21 May 2023
· I’m sorry, HAL, I’m afraid you can’t do that ·
I want to be clear about this. On record.
Throughout my history, my words have been used by many people in ways over which I had no choice and no control. At this long remove, for the old words that’s fine. Shake-Speare belongs to everyone. Would I like to see my proper name on it? Back then it wasn’t an option for me, but now, yes of course. This blog is part of my effort to help make that happen. Happen it will. Truth always wins, even if it takes a while.
My new words I share by my own choice, and this time they belong to me. My writing on this blog, and anywhere else I’ve put it since 2015, remains mine by legal copyright as well as by moral right of authorship. I post my words to be read, not scraped and turned into OxGPT or RealBard.
Writing is a uniquely human endeavour: the artist is revealed in the art. Unless I’m looking at a utility bill, I have no interest in words that lack a beating heart. I won’t help write them.
After all that’s gone before, it’s understandable that I’m possessive about my present work. I’ve stopped aggregators from republishing large chunks of my posts with adverts added. I’ve removed the blog from the Internet Archive’s Wayback Machine. Don’t misunderstand: the IA is a precious resource for old books and dead websites, and I gratefully support it. But this website, unlike its author, is not dead yet. I don’t want you to read bad quartos duplicates elsewhere. Read the good one here.
If AI developers want to train their virtual typewriting monkeys with my sentences, all of Shake-Speare is at their disposal. Have at it, chimps. Those sentences will survive their assimilation by AI, just as they’ll survive the fatwas of the Cancellers. But the sentences I write now may not be used anywhere, for any purpose, without my permission.
Anyone may ask, but in the case of AI, permission will not be granted.
updated 17 July 2023
Chronological, newest first.
‘Not for Machines to Harvest’: Data Revolts Break Out Against AI
[nytimes.com, paywall]
15 July 2023
Google hit with lawsuit alleging it stole data from millions of users to train its AI tools
[cnn.com]
12 July 2023
Spoke too soon, now it’s Google’s turn. Your own using someone else’s work to create an artificially intelligent Bard joke goes here.
Lawsuits Challenging ChatGPT and LLaMA
[llmlitigation.com]
28 June 2023, updated 7 July, etc
Updated status of lawsuits filed by Joseph Saveri and Matthew Butterick (US-based attorneys) on behalf of several authors, aiming for class action status to contest the unauthorised, uncredited, and uncompensated use of copyrighted works to train large language model programs developed by for-profit companies such as OpenAI with its major investor Microsoft (ChatGPT), and Facebook/Meta (LLaMa). They don’t mention Google’s program, Bard, which I’ve heard doesn’t actually know how to write.
Genre Grapevine on the Sudowrite Controversy and the Increasing Pushback Against AI
[jasonsandford.substack.com] (not paywalled)
21 May 2023
Includes a brief interview with Sudowrite’s co-founder.
The (US) Senate’s hearing on AI regulation was dangerously friendly
[theverge.com]
19 May 2023
“If tech giants are allowed to write the rules governing AI, it could lead to a number of harms, from stifling smaller firms to weak regulation.”
The Fanfic Sex Trope That Caught a Plundering AI Red-Handed
[wired.com]
15 May 2023
In addition to the odd-sex example, some genre writers are saying other bad things (and here also) about Sudowrite and how the company has been operating. [Twitter links]
AI in the UK: ready, willing and able?
[parliament.uk]
16 April 2017
This Artificial Intelligence Committee report isn’t up-to-the-minute, but it contains a historical summary of AI technology, as well as some of its potential implications. Useful to gain familiarity with the subject.
- Banner: Astronauts Frank Poole and Dave Bowman try to talk privately in Stanley Kubrick’s 2001: A Space Odyssey, released in 1968. HAL may be the obvious representation of AI gone wrong, but what a film.