On Various New Values Of P(Doom)

On Various New Values Of P(Doom)

Two items over the past week revealed many who are pretty sure the End is Nigh. The Doomsday Clock clicked forward, and certain AI celebrities were caught wringing their hands. Let’s go in reverse order.

Take Roman Yamploskiy. He says he’s 99.999999—that’s a lot of 9s!—percent sure AI is gonna getcha. That’s his value for “P(Doom)“. AI Doom is “existential” risk, the end of Life As We Know It. Yampolskiy is the gloomiest of named persons, followed closely by Eliezer Pudkowsky (here’s an ordered list), who with his mere “>95%” is more like Nassim Taleb at the beginning of the covid panic, screaming at people breathing.

(As I’ve said many times, each time you see “AI” swap in “computer models” and see if you remain as nervous or awed.)

Yudkowsky thinks AI will go renegade and coerce every machine in the world into manufacturing staples. Or maybe it was paperclips. Thumb tacks? Anyway, something to do with stationary. Why and how computer models will do this he never says. But the idea of all these free fasteners has Yudkowsky squealing.

I didn’t go through each man’s specific Doom scenario, but there is rough consensus that to get there AI will stop marching to the rigid beats of hard code and begin to step to—-what, exactly? It’s never really said. The fear is that AI will come alive. Nobody knows how. Nobody even knows how we do. But they say AI can. I say it can’t. (I owe you an argument why not.)

It’s not only people in the AI business who are nervous, but academics outside of it, too, have grown skittish.

How often is a broken clock right? The old joke says twice a day. But if it’s run by concerned people at the Bulletin of Atomic Scientists, it might never be right.

Since 1947 they have rolled their giant novelty Clock in front of cameras and announced, “Look how close the hands are to midnight!” They started out at 7 minutes, or 420 seconds, to Midnight, which they ask us, in the same fashion as AI Experts, to stand for Doom.

Here’s a plot I did of the changes to the Clock (using this data):

Since early this week, we are now only 89 metaphorical seconds before Midnight. One report said:

“In setting the clock closer to midnight, we send a stark signal,” Holz [chair of the organization’s science and security board] said. “Because the world is already perilously closer to the precipice, any move towards midnight should be taken as an indication of extreme danger and an unmistakable warning. Every second of delay in reversing course increases the probability of global disaster.”…

Among the reasons for moving the clock one second closer to midnight, Holz said, were the further increase in nuclear risk, climate change, biological threats, and advances in disruptive technologies like artificial intelligence.

Notice, like with AI, all the Dooms are Science-based.

When the change was announced, I said on Twitter (follow me) “Set it past midnight and see what happens.” I don’t think they laughed.

Anyway, as suspected, Holz means the hands of the novelty Clock to stand in for probability of Doom. How can we transform minutes to probability?

I reason that if the clock is at midnight, doom is upon us, and so the probability of Doom is 1. If the clock is at midnight the previous day, we are in the realm of paradise, and so the probability of Doom is 0. At noon we’re halfway to midnight, so the probability of Doom should be fifty-fifty. Then all we need assume is linearity of time changes mapping to linear probability changes, and we get this Probability Of Doom chart:

Since that’s impossible to read, here’s a blowup (y-axis runs from 0.98 to 1):

The latest change brings us to a probability of Doom of 0.9989699. Not quite as many 9s as Yamploskiy. The Clock started at 0.9951389, and it had a low of 0.9881944 in 1991 when the Bulletin blared “the 40-year-long East-West nuclear arms race is over.”

Two things stand out. The first is that Atomic Scientists are prone to overreaction. The second is that they must be growing more frightened, because the pace of hands meddling has increased and in the direction of Doom. A simple extrapolation says that by about 2040 we’re goners. Which is in line with the AI guys.

Funny thing is these Doomers are the same people asking you to Follow The Science, or they at least argue Science is the solution to what ails us. Science, to them, is thus both salvation and damnation. This is a rich subject we’ll explore more later.

Subscribe or donate to support this site and its wholly independent host using credit card click here. Or use the paid subscription at Substack. Cash App: \$WilliamMBriggs. For Zelle, use my email: matt@wmbriggs.com, and please include yours so I know who to thank. BUY ME A COFFEE.

8 Comments

  1. Cary Cotterman

    I’ve lost count of the times I’ve been told we’re on the brink of doom during my fairly long life. I don’t even hear the noise anymore.

  2. AI, as it is and within presently known technologies,has the potential to answer Fermi’s paradox through an example of one by automating productive work and so producing a world filled with people locked in front of their games consoles whose change horizon consists of asking the machine “what’s for supper?”. So, yes, the doomsters have an argument – I do not believe it personally, but then I’m a starry eyed optimist.

    q.v. Captain Cyborg and the Problem of evil ( https://www.zdnet.com/article/captain-cyborg-and-the-problem-of-evil-6016000489/ )

  3. C-Marie

    Well, you never can tell. I am 81 plus and figure that leaving here will be all Joy as Heaven is where I am going. As long as murdering the unborn and just born for self and science, the Doom is already upon us. How long does anyone think that the only Living God will wait before His final judgment falls upon us. Do repent as needed, if needed, and receive our Father’s gifts of His Loving you and Fathering you, through His only Son. The One World Government and the One World Religion time, is upon us. God is and will be with us through it all. Be faithful to our Father through Jesus Christ by the power of the Holy Spirit.

    God bless, C-Marie

  4. The version of Captain Cyborg on the zdnet site has apparently been somewhat garbled (actually I’m surprised it’s still there). For a clean version – no editorial assistance.. 🙂 ) see my own site: https://winface.com/node/3

  5. You’re right. Substituting “computer model” for “AI” really does defuse the emotion of the language. It almost converts an alarming statement into a head-scratcher. “Why are we worried again?”

  6. The question I have is what is next after the AI bubble pops. Quantum Computing? Maybe nanotech will make a come back. Asteroid mining. Need to get in at the bottom.

  7. McChuck

    The Journal of Anti-American Fraudsters

  8. Barry

    Surely the percentage likelihood of doom must be qualified by a time period.

    1959 had a likelihood of doom of 99.8%.
    Over an hour – didn’t happen
    Over a year – didn’t happen
    Over a decade – didn’t happen
    Over a century – for all of us, certain. For mankind – let’s see…
    Over a million years – yup.

    Obviously worthless without a time qualifier. Wrong with a time qualifier less than a lifetime.

    So ultimately pointless, except as publicity and marketing for the individuals concerned, which it is what it’s all about.

Leave a Reply

Your email address will not be published. Required fields are marked *