Pubblicazioni

2. What this book is and isn’t

Focusing on business risk Focusing on business risk

25/06/2024 25/06/2024

3. What is Generative AI?

Why give the tech industry the benefit of the doubt when they are all but claiming godhood—that they’ve created a new form of life never seen before? Why give the tech industry the benefit of the doubt when they are all but claiming godhood—that they’ve created a new form of life never seen before?

25/06/2024 25/06/2024

4. Strategies for using Generative AI

Sceptical Believers are the ones who take the biggest risks, because they don’t appreciate the pace of change of a paradigm shift Sceptical Believers are the ones who take the biggest risks, because they don’t appreciate the pace of change of a paradigm shift

25/06/2024 25/06/2024

5. Be aware of the abundance of bullshit and snake oil

The AI industry and tech companies in general do not have much historical credibility. The AI industry and tech companies in general do not have much historical credibility.

25/06/2024 25/06/2024

6. Only use specialised Generative AI tools that you’ve vetted

You do not want to become the legal test case for the applicability of a particular law or regulation to AI. You do not want to become the legal test case for the applicability of a particular law or regulation to AI.

25/06/2024 25/06/2024

7. Strengthen your defences against fraud and abuse

The AI provides the method. Bitcoin provides the rewards. The AI provides the method. Bitcoin provides the rewards.

25/06/2024 25/06/2024

8. Prefer internal tools over externally-facing chatbots

You might end up adding novel and innovative security holes to your own products and that’s a kind of innovation nobody wants. You might end up adding novel and innovative security holes to your own products and that’s a kind of innovation nobody wants.

25/06/2024 25/06/2024

9. Prefer transparent and open generative AI

These AI models could be filled with ticking time-bombs. These AI models could be filled with ticking time-bombs.

25/06/2024 25/06/2024

10. Use it primarily for derivation, conversion, and modification

The most productive way to use generative AI is to not use it as generative AI. The most productive way to use generative AI is to not use it as generative AI.

25/06/2024 25/06/2024

11. Are they breaking laws or regulations?

With financial stakes this high, you should only trust what you can verify. With financial stakes this high, you should only trust what you can verify.

25/06/2024 25/06/2024

12. Copyright protections are diminished

Purely AI-generated works provide very little creative or financial benefit in general because of their non-exclusivity. Purely AI-generated works provide very little creative or financial benefit in general because of their non-exclusivity.

25/06/2024 25/06/2024

13. Much of the training data is biased, harmful, or unsafe

You can’t fully trust the output of generative AI because of the massive size of their training data sets. You can’t fully trust the output of generative AI because of the massive size of their training data sets.

25/06/2024 25/06/2024

14. Stochastic plagiarism

AI plagiarism could cost you a fortune. AI plagiarism could cost you a fortune.

25/06/2024 25/06/2024

15. Hallucinations, or AI don’t do facts

Language models are completely unsuitable for use as information systems because of their pervasive ‘hallucinations’. Language models are completely unsuitable for use as information systems because of their pervasive ‘hallucinations’.

25/06/2024 25/06/2024

16. Much of the output is biased, harmful, or unsafe

You’ll sound like a racist granddad who has learned how to write generic LinkedIn posts. You’ll sound like a racist granddad who has learned how to write generic LinkedIn posts.

25/06/2024 25/06/2024

17. Code quality

The cost-saving automation of generative AI is likely to be very expensive in the long run. The cost-saving automation of generative AI is likely to be very expensive in the long run.

25/06/2024 25/06/2024

18. Shortcut reasoning

Their reasoning is nothing more than an intertwined nest of shortcut upon shortcut. Their reasoning is nothing more than an intertwined nest of shortcut upon shortcut.

25/06/2024 25/06/2024

19. Fear of Missing Out

With rare exceptions, there is no first-mover advantage in software. With rare exceptions, there is no first-mover advantage in software.

25/06/2024 25/06/2024

20. The Elegiac Hindsight of Intelligent Machines

Safe, for the tech industry, is too slow. Safe, for the tech industry, is too slow.

25/06/2024 25/06/2024

21. Afterword: a straw to grasp

I’ve never before experienced such a disparity between how much potential I see in a technology and how little I trust those who are making it. I’ve never before experienced such a disparity between how much potential I see in a technology and how little I trust those who are making it.

25/06/2024 25/06/2024

22. Further reading

What to read next after The Intelligence Illusion What to read next after The Intelligence Illusion

25/06/2024 25/06/2024

23. References

The list of references for The Intelligence Illusion The list of references for The Intelligence Illusion

25/06/2024 25/06/2024