Calum Chace's Blog, page 18

Art, creativity, and artificial intelligence

“Art goes beyond creativity, and you have to be conscious to produce it.”

That could be a handy slogan for the forthcoming race against the machines – and it might also be true.

Creativity is the use of imagination or original ideas to create something. Imagination is the faculty of having original ideas, and there seems to be no reason why that requires a conscious mind to be at work. Creativity can simply be the act of combining two existing ideas (perhaps from different domains of expertise) in a novel way.

The eminent 19th-century chemist August Kekule solved the riddle of the molecular structure of benzene while day-dreaming, gazing into a fire. True, he had spent a long time before that pondering the problem, but according to his own account, his conscious mind was not at work when the creative spark ignited. You might argue that Kekule’s sub-conscious was the originator of the insight, and that a sub-conscious can only exist where there is consciousness, but that seems to me an assertion that needs proving.

Can computers be creative? In mid-2015, Google researchers installed a feedback loop in an image recognition neural network, and the result (here) was some fabulously hallucinogenic images. To deny that they were creative is to distort the meaning of the word.

Art is something different. It involves the application of creativity to express something. It might be beauty, emotion, or (my personal favourite) a profound insight into the condition of conscious life. There are probably as many definitions of art as there are people who have sought one, but I suspect that most of us would agree that to qualify as art, a piece of work has to be able to say something to other conscious beings about the experience of the artist. (If that disqualifies a good deal of what is currently sold under the banner of art, then so be it. In fact, three cheers, I reckon.)

To say something about your own experience clearly requires you to have had some experience, and that would seem to require consciousness. Therefore it seems to me that until artificial general intelligence (AGI) arrives, AIs can be creative but not artistic. This in turn means that while Donna Tartt and Kazuo Ishiguro are probably OK for a few decades, Wilbur Smith and James Patterson have chosen the timing of their careers expertly, and their publishers better find something different to do.

Artificial novelists

I just finished reading “The Girl in the Spider’s Web”, the fourth book in the Millennium Trilogy by Stieg Larsson, which was made into a great trilogy of films in its native Sweden, and (IMHO) a less good film by Hollywood. 

Famously, this fourth book was not written by Larsson, who made the ill-advised career move of dying before his books became huge best-sellers. But (IMHO again) it’s damn good. For me at least, the cast of characters live on. (There is an AGI sub-plot which I think is a bit daft, but it barely distracts.)

The shift in author but the consistency of world-building got me thinking about when an AI will first produce a thriller as good as a human. AIs already produce huge numbers of sports reports and financial reports for AP and others. They’re not going to stop there.

Highly successful writers like Wilbur Smith and James Patterson churn out brilliantly readable books, and some of them have stables of ghost-writers doing the actual drafting. There is a formula to these things, and they work. Readers love ’em.

If and when an AI is first able to write a novel as well as a human, will it quickly figure out how to do it much, much better? I can see no reason why not. Thrillers may become like crack cocaine to avid readers: they’ll forget to eat because they are so engrossed in the story.

When? All forecasts are wrong, of course, and I can only offer a science-free hunch, but I reckon within ten years.

That will be another powerful sign that we are well on the way to the economic singularity I describe in “Surviving AI”.

Book launch event for “Surviving AI”

Google kindly offered the use of the conference room at their London Campus for the official launch of “Surviving AI” on 15th September.

David Wood, chair of the London Futurist Group did most of the organisation with his usual efficiency and aplomb, and Kenn Cukier of The Economist did a terrific job of compering.

There was a great turnout and a lively discussion after the talk.  You can watch the video here, and you can join the discussion here.

Interview on Talk Radio Europe

Here’s a link to a 14-minute interview with Bill Padley of Talk Radio Europe, a station for English-speakers in Spain.  A tiny bit of research enabled me to illustrate the power of the AI we already have, much to Bill’s delight.

Review the Future – podcast interview

A couple of days ago I Skyped with Jon Perry, co-founder and co-host of the excellent Review the Future podcast.

(I came across this podcast recently when they interviewed Nikola Danaylov of Singularity 1 on 1, and I’ve listened to some of their back catalogue, which I highly recommend.)

Jon and I cantered over much of the subject area covered by “Surviving AI”, and then he edited it so that I didn’t sound like a complete fool. (Thanks Jon!)

And now it’s online and available for your consideration.

“Surviving AI” is now available at Amazon!

“Surviving AI”, a non-fiction review of the promise and peril of artificial intelligence is now on sale at Amazon.  It’s available in both ebook and paperback formats, and my favourite voice artist, Joe Hempel, is making great strides with the audio version.  (Here are Amazon’s UK and US sites.)

Artificial intelligence is our most powerful technology, and in the coming decades it will change everything in our lives. If we get it right it will make humans almost godlike. If we get it wrong… well, extinction is not the worst possible outcome.

“Surviving AI” is a concise, easy-to-read guide to what’s coming, taking you through technological unemployment (the economic singularity) and the possible creation of a superintelligence (the technological singularity).

Well, that’s what it says on the Amazon site, anyway.

Pandora’s Brain is now available as an audio book

The audible version of Pandora’s Brain is now available at your local Amazon store.  You can even get it free if you subscribe to Audible’s subscription service.

The narrator is Joe Hempel, a very talented voice artist based in the US (see previous post).

In some episodes of Top Gear, the anonymous test driver The Stig listened to audio books while working.  I wonder if he now works for the BBC or for Amazon?  Anyone got his number?

Creating an audio book – by recording artist Joe Hempel

This is a guest post by the voice recording artist Joe Hempel – seen here with a dentist’s best client. Joe has done a great job of voicing Pandora’s Brain, which will shortly be available as an audio book, and he is currently hard at work on the audio version of  Surviving AI.

Joe is a diligent and talented recording artist, and it’s been a pleasure working with him. So I invited him to explain a bit about the process of recording an audio book.

—

When the audition for Pandora’s Brain popped up on ACX (Amazon’s audio creation exchange), I had to jump in.  Books that pique my interest don’t come along very often: the world seems to be filled with sparkly vampires, glorified domestic abuse, and book 75 of series written by ghost authors.  It’s rare that something as smart as Pandora’s Brain comes along, exploring a world that seems almost impossible but is actually not far off.

The challenge with any audio book is to make the characters come alive for the listener.  Hopefully after you listen to the book, you will think I did my job well.  

The creation of an audio book takes a surprising amount of time.  It involves prepping the book, recording, editing, proofing checking, then sending to the author. If the author has any comments then these need to be incorporated and returned. All in all it takes around 5-6 hours for each hour you listen to. 

Pandora’s Brain avoids getting bogged down with technical details. (You can get the whole AI story when Surviving AI, the non-fiction companion, comes out next month.)  In the novel, Calum gives you just what you need to understand the context and then moves on to the action and the character development.  The dialogue is crisp, and the action sequences are fast, action packed and hard-hitting!  There are plenty of twists and turns along the way, and hopefully the audio version conveys the tension and the roller coaster of emotions that the characters go through.

Thought-provoking and exciting, Pandora’s Brain will stay etched in your brain and may prompt you – like me – to go looking for more information on the subject.  It will make you think about your life, your mortality, and what truly makes you…..you!

Book review: The Patient Will See You Now, by Eric Topol

Eric Topol is a techno-optimist and a leading advocate of radical change in the medical profession. Judging by some of the reactions to this best-selling book, parts of that profession see him as a traitor.

Smartphone revolution

One of the two main themes in Topol’s latest book is that the AI embedded in our smartphones is about to spark a revolution in healthcare. Paired with tiny sensing devices and accessing ever-increasing amounts of data, smartphone apps enable patients to monitor every aspect of our bodies.  Diagnoses can be made and prescriptions issued without us having to physically visit the doctor’s surgery, or set foot inside the perilous environments known as hospitals.

Medical paternalism

In other words, Topol is pointing out that the digital revolution which is sweeping across all industries and organisations will lead to big changes in healthcare too. This should probably be an uncontroversial observation, but of course healthcare is different from other industries. There are good reasons why it is conservative and heavily regulated, but Topol’s other big theme is that the healthcare profession is paternalistic and patronising. Doctors are convinced they always know best, and the rest of us are “patients” – people who must wait, rather than “customers” or “clients” who are actively involved in the management of our own health. Generally, we cannot even be trusted to see our own healthcare data, much less direct what is done with it. This attitude, Topol argues, started way back with Hippocrates, and today, he complains, MD stands for Medical Deity; instead of evidence-based medicine we have “eminence-based medicine”.

In every other industry, technology drives down costs and consumers are considered perfectly capable of making decisions for themselves. But many doctors believe that medical information is uniquely hard for ordinary people to understand, and we will become stressed and harder to treat when (not if) we mis-interpret it.

Superior diagnosis

Topol is confident that in time, the clinicians’ resistance to the digital revolution will be swept aside because the benefits of the digital revolution are irresistible. For instance, he points to an impressive range of tests which can be taken with the assistance of smartphones:

“There are now wearable wireless sensors, either commercially available or in clinical development, to capture physiologic data on a smartphone. This includes blood pressure, heart rhythm, respiratory rate, oxygen concentration in the blood, heart rate variability, cardiac output and stroke volume, galvanic skin response, body temperature, eye pressure, blood glucose, brain waves, intracranial pressure, muscle movements, and many other metrics. The microphone of the smartphone can be used to quantify components of lung function and analyze one’s voice to gauge mood or make the diagnosis of Parkinson’s disease or schizophrenia. One’s breath can be digitized to measure a large number of compounds, such as nitric oxide or organic chemicals … Testing saliva, there is the capability to detect strains of the influenza virus and strep throat.”

Testing patients “in the wild”, as Topol puts it, and gathering data (“digital breadcrumbs”) over time, monitoring the impact of drugs, food, exercise, and stress is a better route to diagnosis and prescription than waiting around in the surgery (for an average of 62 minutes in the US, apparently) to get into the exam room and be seen by a doctor for a few minutes. Or rather, not be seen, because the doctor is busy looking at her keyboard to type information into the electronic medical record which we patients are not allowed to see.

Credibility and credit

Topol is carving out a niche for himself as the US’ foremost clinician-turned-revolutionary. He is a practicing cardiologist, and professor of genomics at La Jolla, California. His books have received plaudits from politicians and the media, but brickbats from many of his colleagues. This one is a little rambling and repetitive in places, but it is both authoritative and important.

TL;DR

The go-to book on how AI will impact healthcare.

Humanity’s capacity to believe fiction is our super-power

This may seem a bit off-topic, but bear with me.

Yuval Harari’s book “Sapiens” is brilliant.  I wish I’d written it.  It’s stuffed full of great insights, large and small, and the writing style is crisp, clear and often witty.  You can sample his thinking in this TED talk:

One of Harari’s most important insights is that the reason why humans rule the world, and why the fate of every other species depends on us, is our ability to believe things that aren’t true.  Our ability to believe that money is valuable, and that nations and gods exist is what enables us to organise in flexible ways in large numbers.  That in turn enables us to hunt and kill mammoths, and to build skyscrapers and zoos.

(Social insects organise in large numbers, but not flexibly.  If their current mode of organisation suddenly becomes ineffective, they’re done for.)

So fiction is valuable to us.  Perhaps that is why JK Rowling’s net worth is greater than the price Nikkei is paying for the FT.

And here’s the link to AI – a tenuous one, I admit.  If the ability to believe things that aren’t true is such a vital component of humanity’s use of intelligence to rule the world, will it also be important to a superintelligence?  Will the first superintelligence turn out to be the best novelist (or Haiku writer) that the world has ever known?