Editorial Retort
Is this where we are going? This is not the AI sector that I joined and many others feel the same.
‼️ Trigger warning, but it needs to be said if we are to push for change!
You must have heard by now that a talented young man, an OpenAI researcher-turned-whistleblower, Suchir Balaji, took his own life and was discovered on 26 November after police said they received a call asking officers to check on his wellbeing.
According to a BBC report: "The San Francisco medical examiner's office determined his death to be suicide, and police found no evidence of foul play."
After working at OpenAI for four years as a researcher, Balaji had come to the conclusion that "OpenAI's use of copyrighted data to build ChatGPT violated the law and that technologies like ChatGPT were damaging the internet". He was interviewed by The New York Times (link in the comments).
I have stated this repeatedly in layman's terms and will continue to remind you:
AI is garbage in, garbage out -- or data in, data out.
The idea that AI "generates" or has "creative abilities" is simply *not* true.
Those who try and convince you otherwise are either lying to you in order to sell you something or have no idea how AI works – or both!
Yet, they are still trying to convince you that AI models "do not memorise and regurgitate copyrighted information on which they've been trained"
Well, to honour Balaji's efforts and his name, I strongly recommend that you read his blog post (while it is still available) explaining how to detect and reproduce the copyrighted content trained on by an AI model so you can see for yourself:
https://suchir.net/fair_use.html
I also came across an excellent LinkedIn post (h/t Pascal Hetzscholdt) from Louis Hunt whose team generated some early results of IP they detected in AI models using their algorithms — with the code for recreating them – bad news for tech-bros!
I shared them all below for ease of access, but check Louis’ link for updates. Your mind will be blown!
But my key point is this:
I talk a lot about ethics and responsibility (or the lack of it in the current tech ecosystem), and employees should also be part of this conversation.
I refused to accept or invite external investors or shareholders into my company because the system expects me to put them and their interests above my employees and my customers. Never again!
We must treat our employees as partners, not minions. It's time for the sector to step up.
However, I have no hope for this happening anytime soon.
I cannot begin to imagine the mental pressure this young man was under, and my heart is broken for his family and loved ones.
Please share his link to make sure his efforts are not forgotten. Here it is again: https://suchir.net/fair_use.html
Links to the generated IP findings from Louis Hunt
1 - Here are 383 pages of algorithmically generated verbatims from NYT articles + the code for regenerating them (detected 30,254 articles based on an initial preliminary scan):
https://lnkd.in/eC7bMSc2
2 - Here are 327 pages of algorithmically generated verbatims from Ta-Nehisi Coates, Stephen King, and others' books + the code for regenerating them:
https://lnkd.in/eBhu9F8H
3 - Here are 305 pages of algorithmically generated verbatims from Oxford University associated articles + the code for regenerating them:
https://lnkd.in/e-RB-RJS
4 - Here are 131 pages of algorithmically generated verbatims from song lyrics + the code for regenerating them:
https://lnkd.in/ezn6PYYS
5 - Here are 104 pages of algorithmically generated verbatims from Harvard Business Publishing + the code for regenerating them:
https://lnkd.in/et-cBSzV
Citations:
I see an alarming number of misguided arguments about copyright and AI. Here is my discussion with UK’s leading expert Raoul Lumb for you to understand how this works: https://ceoretort.com/podcast/guest/2024/09/16/raoul-lumb-how-ai-violates-copyright/
Mashable India report (Image source): https://in.mashable.com/tech/86772/what-openai-whistleblower-suchir-balaji-exposed-about-ais-dark-side-before-he-was-found-dead
Latest Episode:
Episode 12 – How a Tech CEO and ex-Pro Athlete Overcame Loneliness and Isolation
Watch or Listen on:
[ YouTube ] [ Apple Podcasts ] [ Spotify ]
Tim El-Sheikh gets really personal and discusses the pivotal moments in his life that led to loneliness and the actionable strategies he used to overcome it. Read more