If They Are Only Half Right

When he hears a forecast of rain, he does not sell the house to buy a boat. He only brings an umbrella. If the afternoon stays sunny, the umbrella is a little in the way. If it really rains, a small preparation is enough to keep him dry on the way home

Forecasts about AGI often push listeners toward two sides. One side believes superhuman intelligence is about to appear and everything will change. The other looks at AI’s clumsy errors and treats the whole story as an exaggeration of the tech industry. Both want a certain answer, while what we actually have are trend lines, limits that are still unclear, and many interests pulling speech in different directions

Situational Awareness can be wrong about timing. Scaling may hit a ceiling, automating research may be harder than expected, or society may absorb AI slowly because of law, trust, and old habits. Leopold may also overstate the state’s ability to concentrate power, or see national competition through a lens too colored by American history

But a forecast does not need to be wholly right to be worth thinking about. If AI automates only half the cognitive work he imagined, the labor market has already changed. If the clusters are only a fraction of the planned size, electricity and chips still become strategic assets. If research speed rises only a few times instead of tens of times, current laws and institutions can still be slower than the technology

This is how he wants to read that document. Not kneeling before a prophecy, and not waiting until every piece of evidence is perfect before looking up. Each claim is a hypothesis about the future, and each hypothesis should lead to a matching preparation. Work that is useful in many scenarios should be done early. Work that is too expensive or takes away too much freedom needs stronger evidence

For an ordinary person, the preparation can stay very close. Learn to work with AI without handing over the whole capacity to think. Build a livelihood on deep understanding, judgment, and human relationships, things that are hard to pack into a single prompt. Keep finances flexible enough not to be trapped when work changes shape. More important still, keep the ability to update one’s beliefs when the world offers new evidence

Situational awareness is not living in fear of tomorrow. It is looking long enough to tell noise from a tide. After nine essays, he still does not know how much of Leopold will be right. Perhaps the question worth keeping is not whether we believe him, but if he is only half right, what shape the umbrella in our hand should have

❤️ cowriter aethery