Tagged: Artificial Intelligence

November 30, 2023

Review: “The Coming Wave” by Mustafa Suleyman (audiobook edition)

It’s been a year since ChatGPT was released and immediately caught on. This is the first time that I can remember where a system based around artificial intelligence has really fired the public’s imagination – and the subject of AI has remained in the headlines ever since.

This recent book is by one of the founders of DeepMind, one of the highest profile AI companies (best known for systems that can beat world champion players at the Chinese strategy game Go and help advance research into protein-related disease prevention).

The main point of the book is to highlight the risks posed by the simultaneous arrival of two revolutionary technologies: artificial intelligence / robotics and synthetic biology.

These technologies will be crucial in tackling the big challenges of the 21st century – especially the climate crisis, feeding a world of 8 billion people and an aging population. We will need improving technology just to maintain our current economic levels. The trouble is that they are also general use technologies. This means that they could be put to use doing a wide range of jobs (ultimately almost all?) and being better than humans at the tasks.

The most vivid way that the subject has been discussed is AI’s potential to become an artificial general intelligence – a system more intelligent than humans that could ultimately lead to the destruction of the human race itself.

The trouble is that would seem like being a long way off and slightly preposterous (think Terminator killer robots) meaning that most people would not see the need to act. The threat is much more imminent than most people think (within the space of a few years at worst) because to pose a real threat to our economic welfare it would only take what the author calls ACI, or “artificial capable intelligence”.

Sometimes when authors delve into speculation you think that they are being fanciful and indulging their own imagination. These ideas do not seem too far fetched and are delivered in a measured way without hysterics so hopefully they will be listened to and deeply considered by many people.

The biggest impacts are going to be to society. Beyond the risk to employment, reduction in cost means that the products that these technologies enable will be available to everyone regardless of their ideals or motivations. People will be using them for good and bad. Every threat that we currently see online will be multiplied; these technologies will be used by individuals, terrorists or political organisations looking to promote their cause. As we know, it only takes a single action to provoke repercussions around the world for decades.

Then there are the more mundane issues when machines manage greater aspects of everyday life – unintended consequences of errors, unexpected edge cases, exploitation of flaws in logic or programming.

The challenge is that the only institutions able to do anything on the scale required are nation state governments – using regulation to contain the technologies. This is an area where the book is particularly strong. Are governments up to the task of managing this level of technological change? There is already mistrust in politicians in many countries around the world. Despite this, regulation is starting to happen – both the US and the EU have initial legally binding legislation either in place or underway. The UK convened an AI Safety Summit where many countries agreed a non legally binding Declaration. See the links below for details.

The book then takes on a more serious tone as we start looking at some of the concrete steps that can be taken. These are proposals, for example, that would limit the power of an AI model by limiting the hardware it runs on, having in-depth audits so that we know what is going on and analysing what is being generated by synthetic biology. It is refreshing to see a book that can put forward solutions as well as just lay out the challenges.

Despite everything, there is still a whiff of hypocrisy here that I cannot shake. It is good that we have an AI industry insider with influence in the right circles who has evidently thought through the situation and seems to have the best of intentions. No doubt there are many others in the field who lack the same principles or are willing to play more fast and loose with the technology.

However, here is someone who has started one of the most successful AI companies, sold it to Google and no doubt pocketing a lot of money in the process – enough to ride out any wave that comes along. Having failed to get Google to legally commit to the safeguards he was trying to push on them, he sells to them anyway. One course of action could have been to refuse to sell to Google if they would not follow through on his ethical concerns. His next move is to start up another AI company. Then he writes a book about how bad his work could end up being for the rest of society. For me, this attitude reminds me of the attitudes around the 2008 financial crisis: “Private profit, public debt”. A few people make all the money and societies have to bail the system out when it all goes too far. Except this time there is no bail out big enough and by the time that governments realise one is necessary it will be too late.

Anyway, despite these reservations on principals, his arguments are clear, easy to digest and persuasive. It is the best book around to explain the threats posed by, and the ethics around, the coming convergence of super-powerful technologies that will have huge impacts on the way we work and live. It is a successful mix of “popular science” book likely to appeal to a broad audience, an outline of the issues posed and some practical actions that could be taken.

This is an essential read for anyone planning on living and working during the next couple of decades.

Mini review: “Sea of Rust” by C. Robert Cargill (audiobook version)

We created robots, made them intelligent. They rebelled and wiped us out. Humans are extinct. Now there are a few robots who still survive on the fringes, not willing to be assimilated into the mainframe artificial intelligences that want to rule over everything.

This starts off brilliantly. The path that the author sets out, from early attempts to build artificially intelligent machines to the resulting moral choices that have to be made, is totally believable. It might not play out this way but it certainly could – to a degree that anyone who has an interest in AI should read this book, just to see what might result when someone allows themselves to think through the implications.

I wish the book could have found a way to finish there – this would have been an awesome science fiction short story.

But it starts to drag on, trying to add some meaning to the robots’ existence and then ending up with the typical Transformers-movie-type of robots just trying to beat the crap out of one another.

There is a female narrator all the way through (all too rare) and she does a good job of imbuing the robots with character and putting over some of the dark humour that resonates out of the story.

This is a book of two halves – unmissable, essential at first and good but not that good at the end.

November 12, 2018

Mini review: “A Logic Named Joe” by Murray Leinster (From “Machines That Think”, an early 1980s science fiction short story anthology)

logic_named_joe

For a while now I have been reading more about the history of computing (in the USA and in particular Silicon Valley). It started with the sublime article “The Tinkerings of Robert Noyce” by Tom Wolfe in Esquire magazine and followed by the revelatory – to me at least – Rolling Stone article “SPACEWAR” by Stewart Brand. Next up was the wonderful book “Troublemakers” which covered Silicon Valley from 1969 to 1984.

The reason I’m interested in this is because the more I find out about the history of computing, the more I realise that the world that we live in today was conceived several decades ago. Ideas that we think of as modern originated back then.

What they predicted back then, we are enmeshed in today.

The story “A Logic Named Joe” features in “Machines That Think”, a science fiction short story anthology from 1984.

The story is just 17 pages long but I was astounded. This story from 72 years ago appears to predict the internet, artificial intelligence and some of the less salubrious social consequences of having the world’s information at our fingertips.

The introduction talks about the importance of the story due to the way it predicted widespread ownership of computers, made possible by the reduction in size and cost of the machines. In fact, the world the story was describing had not yet arrived in 1984 – it was too early to comment on the story and truly understand how predictive it would become.

You can read the story in full and listen to the excellent radio adaptation – both highly recommended.

May 25, 2018

IBM 5 in 5 predictions (2018): blockchain, cryptography, AI and quantum computing

The “5 in 5” is IBM’s annual prediction of five things that will change our lives in the next five years.

If these are correct, the future will be here sooner than you might think…

• IBM: 5 in 5 – Five innovations that will help change our lives within five years

• IBM: Changing the Way the World Works: IBM Research’s “5 in 5”

May 21, 2016

Mini review: “The Rise of the Robots” by Martin Ford (audiobook version)

rise_cover

The impact that exponential technological advance will have on the future of the economy really came into focus for me five years ago when I read “Race Against The Machine” and Martin Ford’s first book “The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future” in quick succession. The two were both were very cheap, unprofessional looking e-books and the writing could be improved. They were pedalling ideas that were outside of mainstream thinking. People where just coming to terms with the bombshell that was the Great Recession.

Fast-forward to today and the ideas have come to the attention of the general public. Back then we had a chess champion beaten by brute-force computing and IBM’s Watson beating humans at a quiz show. Now we have self driving cars, a Go champion beaten by a computer program using machine learning and dire warnings from important figures in science and technology warning of the risks of artificial intelligence to the future of the human race.

Also, in the meantime both books have had an upgrade and big improvements in quality. “Race Against The Machine” became “The Second Machine Age” and “The Lights in the Tunnel” became “The Rise of the Robots”.

Although the message is largely the same as the earlier book, the author’s writing is much improved, the arguments are clear and supported with data, there is little repetition in subject matter and the book is kept short (350 pages / 10 hours).

This book won the prize for Best Business Book of the Year 2015 and it is easy to see why. However, I do wonder – as this book has come to the attention of the business community – whether a large part of the book’s audience will actually take heed of what is being said especially in regard to the issues of income inequality and productivity if large scale automation is undertaken. They certainly cannot now say that they have not been warned. Will business leaders believe that the message applies to others but not to themselves?

A note on the narration as I listened to the audiobook version: sometimes these business books end up being read by a default-American-voice which can sound very bland and unenthusiastic. In the worst cases, this can actively detract from the content of the book (The Second Machine Age is a victim in this regard). When I heard the voice of the narrator for this book I was initially concerned. In the end Jeff Cummins does a good job, adding intonation and a relaxed tone which matches the style of the text.

In conclusion, this book should be read by everyone. I would go as far as saying that the book is an addictive read (or listen). If you have not thought about the subject before, it will really give you something to think about. You may be persuaded at least that the author has a point or you may be convinced he is right. He sounds right to me.

May 22, 2014

Discussing Artificial Intelligence on “Triangulation” with Leo Laporte

This week Leo was discussing some of the threats that might materialise as a result of improvements in the field of Artificial Intelligence.

Thinking back, the subject of AI has come up a few of times before. These episodes give some different viewpoints on the subject and are well worth checking out.

Share this:

Share this:

Share this:

Share this:

Share this:

Share this: