I’m proposing a new Mike Tyson fight to break the internet again

PLUS the product answer, rogue agents and an all bad technology predictions by Techbros edition

PRODUCT ANSWER

Only a couple of votes last week so I will take experimental recipes off of the menu. One person correctly identified that the jacket was a solar powered recharger.

MAIN COURSE & READERS POTLUCK: Robots trying to “Be Like Mike”

Another great reader submission; please keep them coming. Robots in China are learning to box. Although it is comical I find them much better than the half marathoners. Maybe at 70 I am simply amazed at how quickly they can get back to their feet after a knockdown. The running robots weren’t very quick to their after falls and often needed a human hand. Wrestling on the floor with my 5 year old granddaughter usually means 5 minutes to return to the vertical position, so I guess I’m in their class.

I’m more interested in robots now after I heard an AI research podcast that said the different AI approaches being used for robotic intelligence were likely more powerful than brute force LLM scaling, that is reaching its zenith in my opinion. If Trump was really on the ball in protecting American AI interests, he needs to start sponsoring robot contests in every sport right away. The only American robots I have seen are sinister looking security dogs. Let’s see if robots don’t need corked bats to hit a baseball (sorry if you are a baseball fan). China is literally running away in the robot races. The first real robot maid inspired by the Jetsons won’t be Rosie, but Li or Yue.

Late last fall I posted about Mike fighting at 58 in an hugely popular and ridiculous match. I propose a new event when he is 63. Skip the likely unwatchable cage match between Musk and Zuck. People would throw mega dollars down for a Tyson - Robot boxing match on pay per view. The reason I am waiting for 5 years is for the robots to bulk up. Right now they are tiny, some 12 kg (26 pounds for my American readers) below the lowest boxing weight class called strawweight. The match will be a real test for the bots to step up: “Everyone has a plan until they get punched in the face”.

QUICKBYTES : Let your digital agents loose and they just might get captured

I have been pointing out the unmitigated super-hype about agents vs. their actual “binary boots on the ground” impacts. Here is a yet another new worry, which seem to be appearing at a terrific rate.

This concern is similar to the slave computer issue. You know, where you get some malware on your computer which then slows down because it is being remotely used for a variety of nefarious purposes, like denial of service attacks

I’ll just pause here for a moment for people to digest that their actual computers might no longer be exclusively theirs. Many of you just scoffed. The latest estimates are that 20 - 25% of computers are part of a botnet in any one year, so are you sure?

Now your agents can join in the fun and merriment. Here is the ingenious method. As your new digital agent, which is image aware, ventures out onto the internet to book a trip say, it encounters an picture of a sandy beach that has a Malicious Image Patch (MIP) embodied in it. Presto, your agent is no longer yours as it starts to execute this new code that has been implanted in it. Gives a whole new meaning to looking at photos.

Research at Oxford shows that 9 out of 10 agents are hijacked and detecting rogue images is difficult. The code can cause the agent to do anything it has been told to do. So your trip won’t be booked and you don’t really know what your agent did when it was out and about. Like having teenage sons.

A LITTLE SPICE

Human kind cannot bear much reality

TS Eliot

EXTRA HELPINGS - More SpaceX “flights”

Perhaps you saw that Elon’s crew were right back at launching their top of the line rocket called Starship. Another set of problems that resulted in the rocket still not leaving earth but - wait for this extraterrestrial jump - Elon insists that by next year Starship will be ferrying robots to Mars. Did you stay with me on our trip from hard nosed reality to teenage male fantasy?

At least Elon now has a choice between stumbling half marathon robots or very small slow motion boxing robots, as I have’t seen any of SpaceX offerings in public. I’m sure by next year either type will be able to operate completely independently on a different planet.

For my doubting readers, let’s take look at a track record of 5 Elon predictions:

1. 1 Million Robotaxis by 2020: In 2019, Musk said Tesla would have 1 million self-driving robotaxis on the road by 2020. As of 2025, there are none.

2. Manned Mission to Mars by 2025: In 2016, Musk said SpaceX would send humans to Mars by 2025. No such mission has occurred. Wait’ll next year is the old sports cry!

3. Coast-to-Coast Autonomous Drive (2016): Musk claimed Teslas would be able to drive themselves from Los Angeles to New York without human intervention by 2018, which has never happened.

4. Hyperloop Debut in California (2013): Musk predicted a 700 mph Hyperloop train would debut in California within a few years. (No not Fruit Loops and the coloring dye controversy). Project abandoned.

5. Affordable Tesla for the Masses: Musk has promised an affordable, high-volume Tesla for nearly two decades, with repeated claims that a budget model would launch “next year.” Question: How many luxury car companies have become the people’s automobile? None. If you feast on high prices and government incentives you never develop the low price / mass quantity chops.

---

---

MENU MISTAKES

Forecasting is a mug’s game when you have no skills at it. It isn’t shilling for money or recognition

(LEFT to RIGHT: Musk, Geoff Hinton, Reality)

SPECIAL DISH: Change on the Looney Tune Leaderboard.

Most reports of Dario Amodei’s latest prediction acted as if he was quite sensible. (For those not in the know Dario is the CEO of Anthropic, one of the major purveyors of AI models including Claude, which you might remember gave up the secrets to sarin gas despite industry leading “guardrails”). He says that 50% of the entry level white collar jobs will be wiped out in the next 5 years. Well this is a useful stake in the ground that we can now track, just like Elon above. I suspect he is overly optimistic about the AI improvement but he would say that I am one of the people “sugarcoating” labour market realities. (That is his exact phrase).

Fine Dario. But haven’t you also also predicted that cancer will be eliminated in 7 to 12 years. Presumably you realize that cancer is actually over 200 different diseases each likely with different complex causes. As dodgy thinking as your math mistakes.

Haven’t you also predicted that with the untold global wealth AI will be creating, sub-Saharan Africa will grow from their present levels to China’s current GDP level within a decade. Let’s check that thinking:

• The population-weighted average GDP per capita for the 40 plus countries in Sub-Saharan Africa for 2023 is approximately $1,637 (current US dollars).

• China’s is $12,600, current US dollars. So slightly more than 7.5 times large, about 22% real GDP growth per year.

• So AI is the new oil PLUS, as Guyana - with its new oil discoveries - has the best 10 year growth rate in the modern era at just over 10% per year. So more than twice as good across 40 plus nations sustained for 10 years. Ah yes!

Once again Dario’s math is suspect, for a PhD in Physics. Maybe it was actually in Psychics?

There is no way to sugarcoat this Dario but you have taken over the top spot of the looney tunes scoreboard!

WHAT’S REALLY COOKING

Why does the media - whether legacy or social - insist on focusing on technology experts and celebrities for their predictions? They are always terribly wrong. Why do we tolerate this?

Let me introduce Philip Tetlock, who has carved out an acclaimed academic career on the study of forecasting. He is best known for demonstrating that most experts are no better at predicting future events than random chance. He has however found a type of people who are very good at predictions, something he calls superforecasting. Here are the key traits:

• careful probabilistic reasoning,

• constant updating of beliefs,

• intellectual humility

• collaboration in diverse teams.

Sounds exactly like the people in Silly Con Valley, doesn’t it?

Thanks as always for reading. Keep those cards and letters coming in. I greatly appreciate likes, restacks, and especially comments. They keep me youthful.