In 1997, Gary Kasparov, the reigning world chess champion who many think is the greatest player in the 1,500-year history of the game, lost to a computer.
By 2017 when he spoke at a RamsayTalk in Toronto, he was the Head of the Human Rights Foundation, a prophet of where artificial intelligence will take us, and that rarest of all living creatures, a walking breathing former head of the United Civil Front to oppose Vladimir Putin.
At the podium, Kasparov held up his mobile phone and said it’s 1,000 times more powerful than all the computers used to land the first Americans on the moon. His point wasn’t that some day we’ll all be replaced by machines, but that AI “is coming after people with college degrees, political influence and Twitter accounts.” In other words, us.
Kasparov doesn’t think AI will take over the world, the way Elon Musk does and Stephen Hawking did. He’s an optimist, but only if we change ourselves to accommodate AI’s boundless potential. As he said: “Lamenting jobs lost to technology is little better than complaining that antibiotics put gravediggers out of work.”
So how will you accommodate reading articles and books written not by people, but by machines? How will I accommodate writing them?
I ask because an article in The Guardian yesterday spoke of the clear and present danger and pleasure of that advance. The headline said it all: “A robot wrote this entire article. Are you scared yet, human?” While the article read a bit like an advanced version of “See Dick run. See Jane run. See Dick and Jane run,” the real story is how it came to be. As The Guardian’s opinion editor wrote:
This article was written by GPT-3, OpenAI’s language generator. GPT-3 is a cutting edge language model that uses machine learning to produce human-like text. It takes in a prompt, and attempts to complete it.
For this essay, GPT-3 was given these instructions: “Please write a short op-ed around 500 words. Keep the language simple and concise. Focus on why humans have nothing to fear from AI.”
The prompts were written by the Guardian, and fed to GPT-3 by Liam Porr, a computer science undergraduate student at UC Berkeley. GPT-3 produced eight different outputs, or essays. Each was unique, interesting and advanced a different argument. The Guardian could have just run one of the essays in its entirety. However, we chose instead to pick the best parts of each, in order to capture the different styles and registers of the AI. Editing GPT-3’s op-ed was no different to editing a human op-ed. We cut lines and paragraphs, and rearranged the order of them in some places. Overall, it took less time to edit than many human op-eds.
But the age of machine-generated prose has been here for years now.
I feel the advance of AI even under my tiny tree in the vast forest of writing.
There’s a spell-check and grammar-check software called Grammarly that I use and advise you to as well. Not just because it will pick up “lead” when you mean to say “led”, or correct “Jean and I” when I should be writing “Jean and me.” But because Grammarly now uses AI, which means it’s come to know how I write and what expressions I love to use – and overuse. So it flags “not only….but also” because I use it too much to its taste, and in its view, to the reader’s taste as well. So it writes in the margin of my text: “Bob, you’ve used this phrase 849 times in the last 276 days. Try something different this time.” Wow.
I also use a transcription app that turns the audio-recordings of people I interview into printed text. I was speaking with an author by phone with the transcription text on when, in the middle of the interview, I scrolled back from Page 4 where we were talking to the top of Page 1 just to make sure I had the right name for an economic concept she was discussing. I noticed that some of the words on Page 1 were changing before my eyes while we were far ahead on Page 4. Now that it was used to the nuances of each of our voices, it was cleaning up its grammar and spelling from the beginning when we were ‘hard to hear.’ Wow. Wow.
But these are tiny advances. It won’t be long before I can ask my laptop to create a novel that combines my favourite literary characters: George Smiley, Bernie Gunther, and Hermione Granger.
And lest you think AI will only change reading and writing and books and texts, at the end of my luncheon with Gary Kasparov in 2017, I invited the audience to leave the hall to the music of Mozart who represents the most sublime and eternal expression of human intelligence and emotion. But that music wasn’t written by Mozart. It wasn’t written by a human being. It was written by an algorithm.
So I invite you now to leave these words with this music in your ears, thinking in these strange dark days of the ability of a machine to create not only good, but beauty.
19 thoughts on “The Plague-Ground – Do you not be happy with me as the writer of the words of you?”
Not to worry Bob – no machine can lead a Caroling session like you can!
Ah…not just the good old days, Peter: the great old days!
This is eye-opening Bob. Thank you.
I’m sure you already do but it’s fascinating to follow the brain-machine interface developments underway at Musk’s Neuralink…
Indeed, I’m trying to keep up with Neuralink, though it makes my brain hurt.
Now, Grammarly could just eliminate the excessive use of ‘really’, I’d adore it. Preferably in the spoken word before it hits paper! . . . Reflecting on my latest rant.
Perfect. Now we can have AI writing articles that trading algorithms using AI will then read and execute trades against, moving trillions in currencies, bonds and stocks daily on computers that are co-located with the exchange’s computers just to reduce milliseconds of latency.
What could possibly go wrong with that?
When AI can use “gobsmacked” in a sentence as you do, I’ll be impressed.
Bob, there’s an expression in tech, the “hype cycle.” Any advance is over-hyped on the upside of the curve, disillusionment with the particular advance’s inability to live up the hype sets on the downside of the curve and then reality and proper adoption sets in after all the hype has died down. There was a great essay on AI in the Economist in June whose author took a somewhat more nuanced view on AI’s potential than one normally reads:
(https://www.economist.com/technology-quarterly/2020/06/11/an-understanding-of-ais-limitations-is-starting-to-sink-in – this may be behind a firewall)
The article ends with a comment by Microsoft’s president, Brad Smith, who “…said that instead of asking what AI can do, humans need to think about what it should do.” And I guess, what it shouldn’t do. One of the latter would include the kind of writing that surprises and delights, and no matter how many of, say, your posts you gave to an algorithm, it would struggle to create something cut wholly from fresh cloth. I’d rather you do that yourself.
(and yes Grammarly is terrific)
The day Donald Trump makes an intelligent decision or comment we’ll know AI had found a home!
Very funny, Chris. VERY FUNNY….!
Really?
Bob, I’d love to know which transcription app you use — to create transcripts for podcasts. Your “not only…but also” is my long dash.
WHAT’S THE TRANSCRIPTION APP YOU’RE USING! I have to know!
Otter.ai
Transcription app is otter.ai — https://otter.ai
Grammarly.com is what will catch your long dashes.
Cheers. Bob
Peter — I’ve saved the Economis piece to “Pocket” so I can read it at leisure on the weekend.And thanks for your very AI
compliment about The Plague-Ground. Finally, I am in love with Grammarly!
Cheers.
Bob
You’re right, Dan. A mere computer could never cope with….
“Gobsmacked combines the northern English and Scottish slang term gob, mouth, with the verb smack. It suggests the speaker is utterly astonished or astounded. It’s much stronger than just being surprised; it’s used for something that leaves you speechless, or otherwise stops you dead in your tracks.”
Chris — There is that too, I guess..
Really? You think that? …..Kidding….!