Open AI and Anthropic drop new top models

On February 5, both OpenAI and Anthropic released significant model upgrades, within minutes of one another. Both OpenAI’s GPT 5.3 Codex and Anthropic’s Claude Opus 4.6 were optimized for AI software code generation specifically, and for AI Agents in general.

Both models have won widespread praise for their coding abilities, to the extent that the very nature of the job of software development is transforming rapidly from hand coding to supervising a team of AI coding assistants (see story below.)

The “team of semi-autonomous AI agents” architecture is baked in, and is becoming increasingly generalizable to all knowledge work. One can imagine doctors, lawyers, architects, CPAs - really, almost any professional role - outsourcing much of the cognitive work to AI, while the professional focuses on supervising, setting goals for, and validating the work of AI agents. Humans will talk to and build relationships with clients/patients, and take ultimate responsibility for the work, whether done by humans or AI agents. This transformation of the nature of professional work will drive tremendous social and economic change over the next decades.

OpenAI and Anthropic drop new top models

AI Agents are transforming software development - your job is next

With the releases of the latest model upgrades for both OpenAI’s ChatGPT and Anthropic’s Claude (see story above), the AI Agent revolution is moving into full swing. AI no longer just talks to you - now it can do stuff for you, even complicated multi-step tasks that would take you considerable time and effort to do yourself.

The largest and most immediate impact has been on software development. Hiring for junior software developers has slowed considerably, because AI coding agents can now do equivalent work when supervised by senior developers. This represents job security (for now) for senior developers, but their job has changed. Senior developers are hand coding software less and less; now their time is spent mostly in launching and supervising AI coding agents. Because they can launch and supervise multiple agents working on different tasks at the same time, their ability to generate working code soars, but the nature of their work is radically transformed. Computer coding by hand is an example of “deep work”, work that demands intense focus and concentration on a single task for long periods. It can be demanding, but deeply satisfying. Launching, supervising, and validating the output of agents is multi-tasking, and tends to be more disjointed and shallow, which can feel stressful and unsatisfying.

Software is the profession where the AI Agent Revolution is being felt first, but it won’t be the last to undergo a similar transformation. AI agents will likely become prevalent in all knowledge work, and we will all need to become proficient in managing them.

AI Agents are transforming software development - your job is next

ChatGPT rolls out ads, Anthropic snarks

Perhaps you have seen the viral Super Bowl-day ad from Anthropic, in which a 20-something male user is asking AI about how better to communicate with his mother, when the AI abruptly begins pitching him to join a dating site called “Golden Encounters,” where “mature cougars” can hook up with “young cubs.” OpenAI CEO Sam Altman called foul, correctly realizing that Anthropic was humorously slamming Altman’s company for being the first to put ads into ChatGPT responses for Free-tier users, something Anthropic promised never to do with their chatbot, Claude. Facing a projected $155 billion cash deficit over the next 5 years, Altman’s company has a desperate need for revenue, and if people won’t pay, they will have to endure ads tagged to their AI answers. Altman promises that the ads will be clearly marked as such, and that the chatbot answers will not be biased in favor of advertisers.

ChatGPT releases ads, Anthropic snarks

Bezos and other billionaires bet $3+ Billion on longevity startup

Amazon founder and ex-CEO Jeff Bezos is one of the founding investors in Altos, a cell rejuvenation startup that hopes to radically extend the human lifespan and “healthspan.” Launched with a record-breaking $3 billion initial funding and advised by a Nobel Prize-winner, Altos hopes to reprogram human cells to their more youthful forms, essentially turning back the clock on aging. Longevity and rejuvenation are a particular obsession for tech-bro billionaires, when they aren’t renting Venice for the weekend or riding on rockets shaped like a male appendage.

Bezos and other billionaires bet $3+ billion on longevity startup

HBR study says AI doesn’t reduce work, it intensifies it

A recent article in the Harvard Business Review reports on a still-in-progress study of a technology company of about 200 employees as it adopts AI tools. The company funded the AI subscriptions for the employees, who were encouraged, but not required, to use them. For this small group at a single employer, the effect was almost universally the same: Because AI made projects easy to launch, employees launched more projects. Since AI could work on tasks autonomously for long periods before checking back with the user, employees got in the habit of multi-tasking with several simultaneous projects at the same time. Given that AI could fill in any lack of knowledge or skill that a user had, employees launched projects outside of their role - for example product managers and designers began writing computer code. And because prompting AI to begin a task took so little time and effort, employees began to use break time and lunch time to initiate or push forward projects. Over all, the effect was to cause employees to voluntarily intensify their work and to allow work to encroach into non-work hours.

The authors caution that this increased intensity of work is likely unsustainable , and can lead to burnout. Not that anybody is likely to listen until burned-out employees become a problem for the employer.

HBR study says AI doesn’t reduce work, it intensifies it

Tesla’s robot dreams depend on Chinese parts

Elon is desperately trying to pivot Tesla out of the electric car business, where Chinese carmakers are getting ready to eat his lunch, and into robots. Shiny, shiny robots, which nobody actually has, and so are the stuff that dreams (and Elon’s self-promoting hype) are made of. Tesla has already announced that it is discontinuing two of its four car models, supposedly to make space for robot manufacturing. The trouble is, this move only kicks the can down the road for Elon. The major suppliers of robot parts are all Chinese manufacturers. So, Tesla’s US manufacturing plant for robots will only be a final assembly site. And he will be buying his parts from Chinese manufacturers who can make their own robots, or supply local Chinese robot manufacturers, for a final assembly price far below what Elon will need to sell his American-made ones for. Elon is facing the same cost trap he faces in EV’s, just with a different, more hype-able product. Expect Elon to do well as long as hype can outrun reality. Then, not so much.

Tesla’s robot dreams depend on Chinese parts

Chinese Bolt robot runs faster than almost everyone except Usain Bolt

Chinese robot manufacturer MirrorMe has released footage of its Bolt humanoid robot running at a pace of 10 meters per second, or over 22 miles per hour. For comparison, the world’s fastest human, Usain Bolt (likely the inspiration for the robot’s clever name) holds the world’s record for speed for 100 meters - an average speed of 23.35 mph, and a peak speed of 27.8 mph. Hard to believe that just a few months ago, at last summer’s Robot Olympics in Beijing, the winner of the 100-meter dash was a Unitree humanoid that clocked a paltry 10 miles per hour.

Chinese Bolt robot runs faster than almost anyone except Usain Bolt

Epic rolls out AI charting

Industry-leading electronic medical record (EMR) software company Epic has announced that it will be offering AI Scribe and order entry functions in its flagship product. Until now, Epic has allowed a select few AI Scribe companies, including most prominently Microsoft and Abridge, access to its EMR system. These external products have not been granted full integration into the Epic application, generally requiring AI-drafted notes to be copy-pasted into the EMR.

Now Epic is making clear that it is ready to roll out its own fully-integrated AI functions for clinicians, administrators, billers, and patients. This presents a familiar dilemma to Epic clients - opt for an existing best-of-breed solution for a function that may be cheaper, or wait longer and pay more for Epic’s version that offers full integration into the Epic application.

Epic rolls out AI charting

Biocompatible bubble micro-bots seek and destroy cancer cells

CalTech researchers have developed simple biocompatible microrobots that autonomously target malignant tumors and deliver a chemotherapy payload directly, thus avoiding systemic chemotherapy that can ravage the whole body. The researchers developed a system to manufacture a shell of hydrogel that surrounds a bubble of chemotherapy drug. To one end of the shell, they attached chemicals that react with urea, an abundant waste product in the blood, to produce carbon dioxide, which gives the shell a small put steady push in the bloodstream. To the other end, they attached chemicals that react with hydrogen peroxide, which is produced by cancer cells, in a way that exerts a small but steady pull toward the cancerous source. In this way, the bubble-bots “swim” to the tumor. Once congregated around the tumor, the swarm of bubble-bots are burst with a pulse of ultrasound, releasing their chemotherapy payloads simultaneously, creating a lethal environment for the cancer cells while sparing the rest of the body. This method has already successfully treated bladder cancer in mice, and is being further developed for application to humans.

Bubble bots seek and destroy cancer cells