Introduction
Professor Michael Wooldridge (Ashall Professor of the Foundations of Artificial Intelligence at the University of Oxford) explains the difference between AI, Large Language Models (LLMs) and human intelligence. What tech billionaires call AI isn’t that at all. This calls into question many of the assumptions we make about what they can do. What has AI taught us over the last five years? If anything, it has taught us a lot about the importance of nature, is Wooldridge’s view.
‘Human understanding, human experience derives from billions of years of evolution in the natural world and the adaptation of organisms to their environment over that extremely long period. If all AI has told us over the last five years is to be believed, it is that the human understanding, human experience, and human intelligence, are much more complex and nuanced than we thought before we had large language models. And that really is a truth’ – Michael Wooldridge, Faraday Lecture.
Chatbots are not AI, they are Large Language Models
After watching Woodridge’s presentation you should realise we are being sold a pack of lies. For convenience, I have linked here to the last 5 minutes where he sums up his talk. But it’s worth watching all of it, for very simple explanations of why ‘this is not the AI we were promised’.
These are LLMs that cannot meet any test for intelligence; and never will, because of limitations in their basic design. Sure, they sound fancy, spitting out technical words. But they also have qualities that convince humans, on a deeply psychological level, to believe and trust in them. This is dangerous, as it is already causing AI psychosis in at least half a million people.
But all they are is a computer program that gives you the most likely next word, based on training data.
That’s it.
They are not the potentially useful algorithmic AI that Wooldridge and his teams invented.
Because of this, they are irrational. And despite what we have been told, this is dangerously misleading. Oddly, the world takeover isn’t by AI, it’s by the people selling the lies and literally altering hearts and minds.
I have professional colleagues who have suggested how chatbots can learn. Even that’s not true.
‘They hallucinate’ … badly. ‘LLMs are incapable of saying they don’t know the answer to anything’, Wooldridge explains, ‘they are ‘painfully inconsistent’; incapable of differentiating fact, knowledge and beliefs; and don’t know truth from falsity’.
Once an LLM is trained, Wooldridge says, there is very little scope to take back learning data. Continual learning is a very big problem. They cannot revise their beliefs in the world when given new information.
Lessons from nature
The enormous sales and marketing hype surrounding LLMs is owed to the fact that they are financial failures. It’s simply a commercial imperative. Meanwhile, we are wasting resources and killing our mental health.
Wooldridge believes that LLMs have created ‘no net increase in benefit for humanity.
We’re missing a greater opportunity to understand Actual Intelligence … the nature that is all around us. Collectively, we harbour far more intelligence than any machine ever can.
This is why I wrote How to Survive the Next 100 Years: Lessons from Nature. Please, please, if you get the chance to read this, do so. It’s an important book, as it sets out an alternative path that is more fulfilling and proven. It builds human capacity, empowerment and purpose. This is what we evolved on this planet to do.
In all of the noise that surrounds LLM technology, it’s difficult to get the real story about human nature out there.
As a side-note, by the way. AI Overviews on Google Search Engines have now all but destroyed your chance to find my blog online. If you’re reading this after signing up to my mailing list, you may become one of the few people to ever hear about it
The ability for someone like me to reach people like you has just been denied by Google. The same tech firms wanting to sell you their expensive Large Language Models, or force you to pay for more expensive goods, have removed your chance to find meaningful content through searches. Instead, you will only receive the misguiding voice of an LLM.
Large language models cannot work out how to save the planet, rebuild food security, restore water cycles, lower cost of living and enable better mental and physical health.
But we can.
What do we do instead?
The answers to all of this is in how we connect to the natural world and each other.
To enable this better cooperation we need to turn to AI, or the ‘actual intelligence’ of a global network of animals that contains knowledge that artificial intelligence will never know about. Every day 20 quintillion lifeforms make autonomous choices that turn vegetation into habitat. Humanity depends far more on this collective intelligence (that we might call ‘nature’) than our own technology’ – How to Survive the Next 100 Years: Lessons from Nature
The solution is in harnessing the emotional needs of people. It’s in collective decision-making and purpose. If you want to know more about how this is done, then sign up to my mailing list at the footer of this page as I’ll be posting more about this in due course.
Next time you think about using ChatGPT though, remember what Wooldridge says:
‘It doesn’t understand that it is in a world that is changing. … if you leave a conversation with ChatGPT, it isn’t getting bored. LLMs have no conception of the passage of time.
Rational minds are not like that.
Is it sentient? Is it like us? No!’
Maybe instead, read a book, talk to a friend or go for a walk in nature.
