Items I happen to view each day. Science, Technology, Management, Culture, and Writing

    This is my day book for this week. It is a log of things I see on the Internet.

This week: 6-12, April 2026

Summary of this week:

• The Russians are still in Ukraine
• US-Israel-Iran conflict continues
• A two-week ceasefire is created
• Anthropic has a new cybersecurity tool called Glasswing
• Adobe releases an AI tool that creates study aids
• Anthropic's new Mythos may be too good at everything
• The Artemis crew returns safely after one pass around the moon

Monday 6 April 2026

Meanwhile in Japan, they are running out of people. Hence, they are a world leader in physical AI and robotics. The machines are needed to care for the elderly. The same is happening in China. Whoever thought that one day China would be short of people?

Netflix pushes into video AI. Remove an object from a video and the remaining objects interact as if that object were gone. Spooky and quite difficult to accomplish.

Meanwhile in China, lots of folks have drones. The Communist Party, however, is limiting the use of the drones by the subjects of their rule.

Let's all go buy a robot lawnmower. They are pretty good now, so says the article. Perhaps before I die I will have one.

Software bloat comes to the Unix world as Ubuntu now needs more RAM than Windows 11. Horrible.

Ah the journalists are catching on. Let AI "write the news." Basic text processing.

Now playing in theaters with independent distribution is the movie The AI Doc: Or How I Became An Apocaloptimist.

Quoting: Writing at speed, into the dark, without an outline, or any rewriting? That needs training and a slow build up, just as running dow. They are skills you work up to, practice and train for, and then just let go and do it.

Some thoughts on the Pomodoro technique. Work in timed sprints. Rest. Work. Repeat.

Write your memoir. Don't expect much support from the family.

... if you want to watch less television, disconnect the TV after every viewing session. --- Seth Godin

Tuesday 7 April 2026

Meanwhile in professional sports in America, the pro basketball team (men) associated with the University of Michigan won the annual TV tournament. In like manner, the pro basketball team (women) associated with the University of California at Los Angeles won the annual TV tournament.

In Major League Baseball, we are now using an advanced optical system to appeal called balls and strikes. The umpires are usually shown to be correct under the current system that is skewed towards "don't challenge that, stupid." And, if any part of the ball just barely ticks the pane of glass on home plate, it is a strike. Huge advantage to the pitchers over the hitters.

At least Meta is honest and outspoken here: companies are awarding employees for using AI systems the most.

OpenAI funds some studies about safety in AI.

We used to have search engine optimized. Now we have, well, we haven't found a clever name yet but it is something like AI found it optimized.

And in that other thin body of ocean (the Formosa Strait), the communists on the mainland dearly covet the semiconductor manufacturing plants on the island.

Samsung reports huge growth in the first financial quarter. These numbers are hard to believe for such an established company. 8x growth, that is 8x not 8%.

Broadcom will build more of Google's TPUs and rent more compute power to Anthropic.

The New York Times is far behind on this story. Where do they live? The dark side of the moon?

This is seen as some sort of genius management tactic, but good grief, if you pay people more for commuting into the office, they will commute into the office more.

The orbit path of the Artemis mission breaks some record for distance from the earth. Anything for publicity.

Movement in the income distribution in America. Folks interpret the same numbers as great and horrible.

Wednesday 8 April 2026

The money is starting to pour into Anthropic at a rate of $30Billion (with a B) a year.

Artemis flew to the moon, drove around the block, and is heading home. Now comes the real test...re-entry.

It isn't perfect, but the technology has come a long way. See the robot do some actual tasks.

First-world problems for our millennials. Oh woe is them. Really?

More gloom and doom in a remarkable world. You don't like social media? Don't use it, don't pay attention to it, and don't complain because people aren't acting the way you think they should.

Given the low cost of compute power, the much larger number of semiconductor makers, and all that, it is hard to imagine that we are short of computers. Imagine it or not, it is real.

The tech industry is losing jobs. Worldwide, these numbers aren't big, but climbing.

Where the money is going? Cybercrime and scam takes more and more, up 25% last year.

Bill Gates to testify to Congress about the Epstein files. By the time this happens, will anyone be interested?

Warnings that Iranian hackers are attacking US energy and water infrastructure. Better late than never or something. Of course they are doing this.

More adaptive image processing on Android devices. As an old computer vision hacker, nice stuff. The algorithms are not new, but the processing power in your pocket sure is.

More updates in all the big browsers. Remember the browser wars? Remember the word processor wars? Remember... Oh well, now we are in the AI model wars.

And now we have Project Glasswing from Anthropic. It is some type of cybersecurity tool. At least is has a cool name.

Strong rumors that Apple will release a foldable iPhone real soon now (in 2026).

Quoting: The Trump administration is planning to cut the budget of the U.S. Cybersecurity and Infrastructure Security Agency (CISA) by at least $707 million for 2027. When was the last time you heard someone praise the excellence of CISA. The cuts are probably overdue.

Adobe announces an AI tool for students called Student Spaces. Given a PDF document, the tool will generate several types of study aids. Good.

Thursday 9 April 2026

There is some type of ceasefire in the Iran war. Will it last for two weeks as promised? Watch the prices at the gas station.

The casefire may have already been broken. Expect many reports of much cheating.

People are saying, no confirmation of this, that Anthropic's Mythos is such a powerful hacking tool that it is being withheld. It is a cybersecurity tool. It finds holes in security. That is good and that can be bad. Someone now has the keys to the kingdom. Will they share it? With whom will they share it?

A relatively new method of making super-powerful processors for AI is advanced chip packaging. Intel is pushing into that packaging as fast as it can.

An in-depth description of one group's journey to storing data in S3 files on AWS.

Building blocks for software. Everything old is new again. Good stuff, but please, let's not act as if this is something new.

This piece describes that value that some people can bring in our current era of text extruding machines. The writer calls it good taste or distinction under uncertainty.

A look into the history of OpenAI. The piece characterizes OpenAI as not being worthy of anyone's trust on any matter. The old Google phrase of Don't be evil comes to mind.

The Rivian R2, coming real soon now, is rated at 335 miles of range. You can drive your $60,000 vehicle for five hours and then wait for two hours before you can drive again. Makes sense?

Meanwhile in the UK, OpenAI pauses a big-money project due to costs and regulations. The cost of doing business in some places is just too high. Too bad for jobs in those places.

YouTube has a new Shorts feature. AI will create an avatar of me, use my voice and face, and make an eight-second video. Great fun.

We can now search ChatGPT to find videos on Tubi. Again, great fun. I suppose this solves a problem for someone. It seems there are better uses for all this technology.

Quoting the headline: The OpenAI Foundation says it is working to finalize over $100M in grants this month, across six institutions, to support and accelerate Alzheimer's research. This is something useful to apply our technology.

We are beginning to come out of the chaos phase of change and settle into a new status quo.

We still use undersea cables to carry the great majority of Internet and other data. These cables are just lying there available to any ne'er-do-well to damage. Companies are finally working on tools to protect the cables.

This one will take some study and practice. Google puts a new Notebooks feature in Gemini. This somehow links to NotebookLM and enables a person to do all sorts of new wonderful things.

A Federal court in Washington D.C. reverses the decision of a Federal court in California regarding Anthropic and the Department of Defense or War. Predictable and predicted. Courts in different backyards see things differently. Culture clash.

Quoting another headline: A hacker claims to have stolen 10PB+ of data, including classified defense docs and missile schematics, from China's National Supercomputing Center in Tianjin. Don't ya' just love the hacktivists?

Patreon claims success with millions of paid users and big gains in revenue.

X now has translation worldwide. Do we really want everyone to know what everyone thinks?

ICE reports 100% gain in arrests. They are arresting criminals, i.e., people who break the law. Some of these laws may be misguided and unjust. There are ways to change those laws. And sometimes, well, law enforcement abuses its power. Still, this is a success. Law enforcement is enforcing laws more than before. That is their job.

Friday 10 April 2026

Excellent piece on why Agile failed and why AI can succeed. A new tool requires a new process. It just happens that the new process resembles the old waterfall.

Quoting the headline: Meta's Superintelligence Lab unveils its first public model, Muse Spark

Spending 17 years trying to solve the great Bitcoin mystery. I guess there are worse things to do in your spare time.

How one company had everyone using AI tools.

Think a little deeper about Anthropic's Mythos. It is good at cyber because it is good at everything. This piece seriously pushes the idea that it will replace white-collar jobs. And that can be good. On the hother hand...

Quoting: The CIA is starting to deploy artificial intelligence to boost one of its most hallowed missions: analyzing the plans, intentions and capabilities of foreign nations

TSMC reports a big financial quarter. It is amazing to see an established company with this size growth.

When state-by-state AI regulation is to the advantage of the bigger AI companies. Work each state one-by-one and have favorable bits and pieces. Smaller companies cannot do this as they don't have enough lawyers.

The other side of state-by-state regulation: each state tries to grab some of the money.

Join a union, lose your job. Apple closes a retail store. The union sues. The idea that a union or a judge can require a business to keep a money-losing store open is preposterous, but in this day and age, why not?

Oh, by the way, despite all the technical success of SpaceX, they lost $5Billion (with a B) last year.

Quoting the headline: A gap in understanding AI is growing, as casual users cite flaws in old free models while power users point to new models' staggering gains in technical domains. The situation is changing rapidly. Learn. Adapt. Learn. Adapt. Repeat almost hourly. Long-term strategies and studies are of little use.

Pushing and shoving continues between OpenAI and Anthropic. Perhaps the world can have two dominant companies in one market. Coke and Pepsi continue to battle (sort of). Can this happen in AI?

And Meta is trying to join the fray.

The Electronic Freedom Foundation (EFF) closes its X (Twitter) account. People simply are not reading what is posted on X. EFF will stay on the other big-name sites even though it doesn't endorse all their practices.

OpenAI starts a $100 per month subscription. It appears that AI tools are about to go the way of Microsoft's Office 365 (or whatever Microsoft calls it these days).

Anthropic bumps up what is available to paid subscribers.

America's 20 somethings are using AI, but pessimistic about the whole situation.

Saturday 11 April 2026

The Artemis crew returns safely.

Rumors that Tesla will produce a smaller and less-expensive car next year. These will be built in China.

Build your own AI infrastructure and such so that all employees will use AI without being AI experts.

I like this piece. It is short, but quite good. It explains how if the last time you used an AI system was six months ago. You probably have a completely wrong impression of what they can do. The old question, "Should I drive or walk to the car wash?" is completely gone.

Without significant change in lowering the electric power required of super computing power, we won't have any super duper computers in the home.

This person has a firm grasp of the obvious, and that is pretty rare these days. To use these super powerful AI systems, you need to frist know what you are doing.

AI and golf. Golf is a business. The current slate of AI systems do basic business operations.

Anthropic is slowly but surely moving into business operations. OpenAI was already there and is holding steady in the market. The trend, however, is in Anthropic's favor. Now if Anthropic's leaders would sit down with our government and hold the ego in check long enough to settle the spat.

Nice try by some folks working for our government in civil service. They are too far behind and move too slowly to matter. The Mythos events of the week show that.

I want my bank to use Mythos!

Finding a faster way to build datacenters. Okay, let's do it.

Is it a vaccine or a therapy? Good grief. Just do the work and help people. What's your problem?

Scientists discover that chimpanzees have civil wars. In plain language, someone happened to see and document how chimps kill each other. Animals fight to the death for resources. Nothing new here. Since these are apes, we someone expect something more civilized. Silly notion.

Surveillance in Mexico. Someone has bought and integrated the technology there.

The conflict with Iran has shown the obvious: datacenters are big and easy to destroy. Folks are now looking to the more obvious how distributed systems are more resilient. Then there are data embassies. Folks, even in wars, tend to avoid destroying embassies. Make a big embassy grounds and put the datacenter inside the walls.

Sunday 12 April 2026

I like this piece about how one company moved to AI-first work. Whatever your call it, they changed how they work. People went along with the change. Lessons here for many groups of people want to do things differently.

Some technologists from Anthropic meet with "Christian leaders" to discuss ethics and such in AI. No list of who these leaders were or why they were chosen for the meeting.

Ethical AI? Anthropic is the worst by a wide margin, with a crawl-to-refer ratio of 8,800 to 1.

There is some sort of brain drain in the AI research community as some are returning to China for better pay and better quality of life. Silicon Valley doesn't have those things? Odd.

Meanwhile in the Netherlands, they approve Teslas self-driving system or sort of do that. I don't understand what supervised self driving means.

Buying choices: MacBook Neo or Air. The first question to the it-depends answer is, is this for a job you do 8 hours a day? The answer is Air.

And now we have the electric-powered cargo bicycles. These things still cost too much. I guess the price will never come down to $500 where it should be. I am an old man and don't understand why you pay $3,000 for a bicycle that has a little cheap electric motor.

Like to dissemble your laptop (who does that?)? Don't buy Apple or Lenovo.

Success leading to failure: Apple built its MacBook Neo computers using leftover processors. Yes, they have run out of parts. And now they have to do something that isn't economical.

A big drop in births to teenage women is good (I think so). The big drop is US birthrates overall is not good (again, I think so).

Per this study, AI agent skills (too many adjectives here) fail in the real world.

Quoting the headline: From GPT-2 to Claude Mythos: The return of AI models deemed too dangerous to release