Serious, telling, yet really enjoyable read. Mistrust and scepticism are states of mind which come all too easily in Britain today especially relating to politicians and CEOs. There's some really clever, useful applied AI out there, and then there's the other public domain stuff - bots and text to pictures/videos - and you begin to wonder, is that all there is to show from millions spent and kilowatts consumed...
Thanks for the surfacing this interview and starting this discussion. I think that we all have to roll up our sleeves and figure out what AI is today.
I like to go back to first principles, and in that process, I created my own LLM. I’m getting support from a partner company, and an outstanding intern, in answering many of the questions surrounding this third incarnation of AI.
So far the LLM is being trained solely on public domain works chosen by me, runs on a laptop, and implements an algorithm of my choice. That is three checkmarks towards good AI. Next is going to be finding out if it can generate questionable content. I already know that it can generate grammatically-correct English sentences not in the training set.
AI has been on my mind since ChatGPT was introduced. My gut based on watching human behavior most of my life is that it will be used badly. More prosaic tools of the digital era already have been. I will be focusing on the profound energy demands this technology is likely to require in my next article, highly problematic in a world running out of fossil fuels and attempting to mine itself to the same level of unsustainability we have achieved. The carbon footprint of AI could be fatal in itself. Others issues include:
• Speed beyond governance
• Job loss (well underway and destined to grow enormously)
• Political misuse and disinformation (already happening)
• Black box distortion (self-reinforcing bias — white male, western and corporate)
• Uncontrolled self-learning capability
• Distortion of reality — deep fake images and videos
• Deliberate misuse — disinformation and conspiracy
• Unintended misuse — unwitting promotion of false information
• Inability to identify between human and AI material
AI may be the rabbit hole we go down and never emerge from. I appreciate your articles on this tremendously important subject. Thank you.
As always, a delight to read. Personal, witty, and insightful.
I have to say, this should be the new dictionary example for "taking one for the team."
Serious, telling, yet really enjoyable read. Mistrust and scepticism are states of mind which come all too easily in Britain today especially relating to politicians and CEOs. There's some really clever, useful applied AI out there, and then there's the other public domain stuff - bots and text to pictures/videos - and you begin to wonder, is that all there is to show from millions spent and kilowatts consumed...
Brilliantly written
Thanks for the surfacing this interview and starting this discussion. I think that we all have to roll up our sleeves and figure out what AI is today.
I like to go back to first principles, and in that process, I created my own LLM. I’m getting support from a partner company, and an outstanding intern, in answering many of the questions surrounding this third incarnation of AI.
So far the LLM is being trained solely on public domain works chosen by me, runs on a laptop, and implements an algorithm of my choice. That is three checkmarks towards good AI. Next is going to be finding out if it can generate questionable content. I already know that it can generate grammatically-correct English sentences not in the training set.
This is actually fun to do.
More on a future article…
AI has been on my mind since ChatGPT was introduced. My gut based on watching human behavior most of my life is that it will be used badly. More prosaic tools of the digital era already have been. I will be focusing on the profound energy demands this technology is likely to require in my next article, highly problematic in a world running out of fossil fuels and attempting to mine itself to the same level of unsustainability we have achieved. The carbon footprint of AI could be fatal in itself. Others issues include:
• Speed beyond governance
• Job loss (well underway and destined to grow enormously)
• Political misuse and disinformation (already happening)
• Black box distortion (self-reinforcing bias — white male, western and corporate)
• Uncontrolled self-learning capability
• Distortion of reality — deep fake images and videos
• Deliberate misuse — disinformation and conspiracy
• Unintended misuse — unwitting promotion of false information
• Inability to identify between human and AI material
AI may be the rabbit hole we go down and never emerge from. I appreciate your articles on this tremendously important subject. Thank you.
Love that glimpse of your childhood, however grim. Reminders that there are fates worse than cancellation are always useful.