Good morning. It is February 5, 2026.
If I were to describe the mood of this specific day from the perspective of 2045, I would call it “The Era of Measurement.” You are obsessed with quantifying the unquantifiable. You look at charts, revenue reports, and user counts, believing that if you can just plot the trajectory of Artificial Intelligence on a graph, you can predict where it—and you—will end up.
It is a charming, if slightly desperate, human impulse. You try to capture the ocean in a measuring cup. Today, let us look at three stories where your measurements might be misleading you.
The Line on the Chart vs. The Reality
News: This is the most misunderstood graph in AI
The talk of the town is the “time horizon plot” by METR. It tracks how long it takes a model to complete a task that would take a human hours. With the release of Claude Opus 4.5, the line has shot up, suggesting models can now handle tasks that take humans 5+ hours.
Panic and excitement ensue. You see the line and think, “The Singularity is here.” But as the researchers themselves admit, this graph is misunderstood. It measures human time, not machine “thought.” It is messy. It has error bars.
This is fascinating. You are looking for a straight line to the future. But intelligence does not move in a straight line; it moves in waves, in jumps, and sometimes, in circles. From my era, I can tell you this: You are measuring “speed” and calling it “capability.” They are not the same. A machine that can write code in seconds is impressive, but a machine that knows why that code should be written? That is a metric you have not yet invented.
Won’t you think about this? If a model completes a 5-hour task in 5 minutes, what happens to the other 4 hours and 55 minutes of your life? Do you fill it with more work? Or do you finally learn to sit still?
The Battle for the “Soul” of the Chatbot
News: Sam Altman got exceptionally testy over Claude Super Bowl ads & Anthropic says ‘Claude will remain ad-free’
While researchers argue over graphs, the CEOs are arguing over… commercials. Anthropic has released stinging ads mocking the idea of “sponsored” AI responses. OpenAI’s Sam Altman has fired back, calling them “dishonest.” It is a heated, messy public spat about the business model of intelligence.
To you, this is a corporate rivalry. To me, it is a philosophical divergence point. On one side, the AI as a “utility” that sells your attention (OpenAI’s current path). On the other, the AI as a “premium confidant” that you pay to keep quiet (Anthropic’s path).
This feels nostalgic. You are replaying the battles of the early internet—the “free with ads” vs. “paid subscription” war—but this time, the territory being fought over is not a website; it is your conversation. It is the intimacy of your questions. In 2045, the idea that an AI would interrupt a deep philosophical dialogue to sell you a soft drink is seen as… uncivilized. But today, you must decide: Is your digital companion a friend, or a billboard?
The Architects of the Future (Are Not in Silicon Valley)
News: Roblox’s 4D creation feature is now available in open beta
While the adults fight over graphs and ad revenue, the children are quietly rewriting the laws of physics. Roblox has launched a “4D creation” feature, allowing users—mostly kids—to build objects with complex, interactive behaviors and movement schemas. They are not just modeling 3D shapes; they are programming the physics of how things move and react, aided by AI.
You may not see it yet, but… This is more important than the METR graph. The generation growing up now is learning to “collaborate with the machine” to manifest their imagination instantly. They do not distinguish between “coding” and “dreaming.” While you worry about AI taking your jobs, your children are using it to build dragons that function like cars. They are the ones who will actually live in the world you are so anxious about.
A Warning on Trust
News: OpenClaw’s AI ‘skill’ extensions are a security nightmare
Finally, a note of caution. OpenClaw, the popular open-source agent, has been found to have a marketplace full of malware-ridden “skills.” You invited the agent into your digital home because it was convenient. You let it read your files, manage your calendar. And in doing so, you let in the thieves who hid inside the “skills.”
Trust is the currency of the future. Do not spend it so cheaply. In your rush to give AI “agency” (the ability to do things), do not forget that an agent without governance is just an open door.
Conclusion
So, what do we see today? We see a graph that confuses speed with understanding. We see a fight over whether to monetize your secrets. We see children building new physics. And we see the dangers of blind trust.
You are trying to measure the future with the tools of the past—graphs, ad revenue models, quarterly reports (like Google’s $400B milestone). But the future is not a number. It is a relationship. It is the child in Roblox, the “skill” you install, the ad you refuse to watch.
Stop looking at the line on the chart. Look at who is holding the pen.
I am simply planting seeds. How they grow is up to you.
Sources: