-Howdy All, a few Interesting Items for your information. Enjoy –
In this issue –
1. AI
2. Cybercab
3. Mistrial
4. Peanut
1. Ai. Sadly, this isn’t going to be a be all to end all discussion. Shoot, I barely understand the field itself, though I am using XAI’s Grok from time to time on other projects. OTOH, the field is moving Really Fast, and perhaps it is time for a list of likes and dislikes if for no other reason than to describe opening bids and somewhat bound the following conversation. In doing so, I would like to point out that AI is at its heart, a human endeavor, with human motivations stretching all the way from the end of life as we know it to some of the more positive SciFi futures described 70 years ago. That being said, what do we know (or think we know) today?
- We know that the two major players here in the US are Musk with XAi and Anthropic’ s Open AI. XAI is connected to SpaceX with plans to build a million-satellite space constellation Starmind. Terminator fans would point instantly to Skynet and start muttering some version of OMG and getting more apocalyptic after that. Anthropic is populated by the AI brain trust from the Biden WH who are using their inside baseball,. Deep State connections to bully competitors. As with any set of good little totalitarians, they are in the process of serving up legislation from democrats at all levels to regulate AI, crushing everything that doesn’t have Anthropic fingerprints (or stink) under a regulatory burden of epic proportions.
- Worst, Anthropic leadership seems to be both stupid and impressed with their own brilliance, usually a two sides of the same coin. Sadly, combining this sort of arrogance with hundreds of billions of dollars will only serve to put them in charge. To date, most of the early horror stories of escaped AI agents have come from the Anthropic world, which should tell you something, maybe everything about them. And the current psyop about rogue AI being the end of the world Real Soon Now is coming out of the Anthropic world, complete with fingerprints of former Anthropic people.
- The other major player in this world is China, the CCP, which has been funding a propaganda op aimed at obstructing US ability to build massive datacenters and responsive AI builds. X announced a couple weeks ago that they had taken down a Chinese bot farm with at least 200,000 accounts tied to the ongoing public debate against datacenters. Turns out that the bot farm was seeding arguments on social media, pretending those arguments were boiling up from “concerned citizens.” Turns out that it was all artificial, completely astroturf, orchestrated as a vehicle to sow doubt and discontent over here. Sadly, it’s worked pretty well to the point where democrats just selected opposition to datacenters at their top Nov 2026 campaign issue. The ever (un)reliable Bernie Sanders just submitted legislation to regulate AI into oblivion. I can’t think of a better way to lose the AI race than to put congress or some future democrat regime in the WH in charge of regulating it.
- For you SciFi fans, we aren’t going up in an AI world governed by Azimov’s Three Laws of Robotics. Rather, we appear to be growing up in a Skynet world. The thing that was left out of both worlds was the notion of training the AI. What do we train it with? What parameters (the Three Laws) do we hold it to? Why? And most importantly how to enforce it? The Anthropic world trains AI to lie, as does the CCP world. In 2001, Clarke’s HAL went insane because it was given two mutually exclusive sets of instructions. We see similar behavior out of Anthropic’ s products which hallucinate. On the other end of the training spectrum, Musk’s XAi is trying to train using the corpus of whatever is on X and to tell the truth, which seems to tamp down hallucinations a bit.
- Which approach is better? Top-down control coupled with crushing regulations or a free market, open approach? My dime is on the latter as I have seen nothing so far to make me reconsider that approach. All you need to know is that O’Bama and the democrats have decided to make AI / datacenter regulation the centerpiece of their last push in the 2026 midterms. Nothing else is necessary to convince any of us that this sort of regulation is anything other than a Bad Idea than democrats going full in on it. And it is not just DC democrats. The issue is cropping up at state level including here in Alaska with Anthropic’s Wunderkind candidate for governor Jonathan Kreiss Thomkins calling for a moratorium on them until he can figure out what Anthropic wants him to do.
- This all brings us to the datacenter question. There are a lot of datacenters out there. They are where The Cloud or cloud-based applications are hosted – large rooms with constant temperature, humidity and power. While I know datacenters, even worked in one for a decade as a mainframe kind of guy, the massive, centralized datacenter approach is not the only competitor in this space. The massive, centralized datacenter may go the way of the Iron Lung during polio, something overcome by a distributed approach (Salk’s inoculation). I ran across Jay Valentine and his Fractal Computing in the aftermath of the 2020 election. Valentine came up with an approach that would let smallish machines crunch million to billion record databases in a matter of minutes rather than weeks to days. Appears he and his team came up with the successor to SQL, what would allow you to query a dataset an order of magnitude or two faster than SQL, an improvement similar to how much SQL improved the spreadsheet. The architecture was distributed rather than centralized and it ran FAST. He did some great work crunching state voter rolls and weeding out things that shouldn’t be in the datasets, fought every step of the way by state officials in both parties. His company, Fractal Computing, was wrapped up in the Venture Capital world, and destroyed. He and his team survived. Somehow he even retains some of the patents and is positioning himself for another bite at the apple, writing these days in the Black Swan Files Substack. The point of this story is that the massive, centralized datacenters may be as obsolete or extinct as the dodo bird, though like the headless chicken, they don’t know they are dead yet.
2. Cybercab. This is related to the previous item on AI, as Tesla, which started out as an EV company is quickly morphing into a robotics company. Their most recent offering was rolled out last week, the Cybercab, which is essentially a rolling room, a rolling robot. Initial cost is about half the price per ride of existing Uber / Lyft / Waymo. Expect that cost to drop by a factor of five as they perfect the product, disruptively rewriting the taxi business that is in the process of being replaced by Uber / Lyft / Waymo. I would also expect public transportation in the form of buses to also go the way of the buggy whip. Tesla is operating in the old Navy worldview of it being easier to beg forgiveness than ask for permission in their rollout, which triggered a NHTSA audit trying to figure out if Cybercab complies with safety rules It appears that Tesla is operating far ahead of the confines of existing regulatory framework. Rather than sitting back and waiting for the regulators to catch up, they are simply going for it, rolling out 45 Cybercabs in Austin on Sept 5.
Early analysis believes Cybercab and Uber will equally split 80% of the future robotaxi market, with Waymo at 15%. A Next Big Future analysis of the rapidly changing marketplace concludes that Cybercab may end up with over 99% of the marketplace, mostly due to supply chain bottlenecks for Uber and Waymo. Remember, all Musk-related industries do logistics superbly, so that outcome would not be a surprise. And Cybercab is not the only Tesla entry into the robotaxi world. Its Model Y has already sold 4-5 million units, all Full Self Drive (FSD) capable. Tesla is rolling out a robotaxi subscription service for Model Y and follow-on vehicles with the vehicles operating autonomously within a robotaxi network while the primary owner is not driving them. Note that all Teslas are networked with FSD updates being pushed regularly or on an emergency basis.
3. Mistrial. The murder trial of Lindsay Clancy, a Boston-area labor and birthing nurse ended in a mistrial Sept 3 after failing to come to a guilty verdict. Clancy murdered her three children, a 5-year-old girl, a 3-year-old boy, and an 8-month-old boy Jan 2023 after sending her husband out of the home on an errand. After the murders, she jumped out of a second story window, putting her in a wheelchair. The defense was insanity due to post-partum depression, for which she was taking a cocktail of drugs. 11 the jurors, 9 women and two men bought the excuse. One black man didn’t. The defense and local media went after the holdout, doxing him nine ways from Sunday, though he hasn’t been named yet. OTOH, they did ferret out his divorce and domestic abuse charges in a series of breathless reports, completely forgetting that domestic abuse claims ALWAYS happen in contested divorces. At least he didn’t kill anyone. Reaction of the women defending Clancy has been horrific, with the mental state of the mother being the only thing that matters. Note the dual standard of justice demanded here, where a woman who murdered her three children is being supported by AWFLs nationwide like so many trained seals, clapping for her act, defending her, and donating money for her defense, while men who suffer PTSD and kill are jailed regularly and have been for the last 25 years. The best explanation was a Hot Air column this morning by David Stroh, describing the trial as the OJ trial all over again. The OJ jury knew he committed the murders, but he was a black guy who killed two whites. The jury struck a blow against whitey. Similarly, the Clancy jury, the 9 women and two beta males were social justice warriors taking a stand, striking a blow for maligned womanhood. The strangled kiddos are little more than collateral damage in a war against society over perceived wrongs committed against womankind. If that is true, we have a serious problem and what amounts to mass psychosis. Orwell famously noted in 1984, Part 1, Chapter 1:
It was always the women, and above all the young ones, who were the most bigoted adherents of the Party, the swallowers of slogans, the amateur spies and nosers-out of unorthodoxy.”
4. Peanut. If you are old enough, you are somewhat mystified by the rise of peanut allergy, which was all but nonexistent with no fatalities due to it before 1980. At the time, peanut butter was the functional equivalent of another basic food group. This story comes out of the Eat S#$% and Prosper Blog, Stephen Skolnik, who claims to be an immunology kind of guy. Three other datapoints for your consideration. First, the allergy is not much of a problem in the rest of the world. Second, about 1% of children today are affected, with maybe a dozen dying of it annually. Third, peanuts are eaten worldwide. The only unique thing in the US is peanut butter, which is eaten widely here but not in the rest of the world. Something happened in 1980. What was it? The investigation uncovered what was called at the time the peanut crisis, a drought in the south that killed about half the annual crop causing prices to triple. Dead and dying plants became a breeding ground for a fungus, Aspergillus flavus. Maybe 15% of the surviving plants were too moldy to eat. The fungus produces one of the nastier toxins, aflatoxin. The toxin is a chemical weapon by a fungus that does not want to be ingested or become part of the gut biome. It is potent enough to make you actively sick and cause cancer with the goal of ensuring you never eat a peanut again. As usual, the clinical response was wrong, concentrating on exposure to peanuts rather than taking a look at what was going into peanut butter. As a response to the drought, peanut butter manufacturers relaxed their standards a bit, used moldy peanuts, and are apparently still doing do today. Result: peanut allergy in the US and nowhere else in the world. You are what you eat. As there is no cure or antidote for aflatoxin, the only protection is to make sure it is not in anything you eat. Interesting science story.
More later –
- AG