Love it! I think the power of LLMs to acquire new skills and deepen the knowdledge is underestimated.

When used correctly, they offer a huge advantage over those who don't use them and think they understand but remain superficial. I encourage you to ask even the most obvious questions.

This assumes that “more intelligence” is the objective. Higher intelligence does not imply better values, and even “better values” depends on whose moral framework gets encoded.

Intelligence alone does not, but when combined with knowledge I think it may do.

Having a wide breadth of knowledge and the ability to consider it tends to provide a good foundation of perspectives for positive values.

Most harms that people do are through some form of ignorance. Either by not comprehending the consequences of their actions or by not knowing better ways to express themselves.

Geoffrey Hinton mentions that Humans are more intelligent than animals, and we keep animals in cages. The counterpoint is that the intelligent people who know the most about those animals are the ones who work the hardest to have those animals in cages moved to more suitable environments.

Congrats! Most people won’t want to debug drivers, kernels, ACPI, adapters, and fan headers. But for those who do, the capability-per-pound is absurd.

No, there is nothing new in this article. This methodology is already adopted - and further optimized - by the sw community.

> In the context of our study, we did not observe significant differences in trust between the free and the paid versions, however, 2 participants mentioned trusting Gemini Advanced more because it is the paid version. Instead of depending on AI-generated code for security measures, developers might use AI as an extra layer of support, particularly for identifying potential vulnerabilities or addressing security measures that may have been overlooked.

In your experience, is AI useful as an extra reviewer or as "authority"?

Most people think the current valuation is for the models themselves. Actually, they're building the infrastructure for the next 50 years.

> Actually, they're building the infrastructure for the next 50 years.

What infrastructure? The hardware would be outdated in 3 - 5 years, after all. What other infrastructure is needed for AI?

The dark fiber of our time?

I'm experiencing the same. Codex gtp-5.5 has more brilliant intuitions, write less code, i.e. it identifies the exact point in which the modification shall be done. Nevertheless, huge improvements on personality from opus 4.7 (it was too accomodating) to opus 4.8

I don’t think meetups are dead, but generic meetups feel much weaker than before. I think the hard part for young people is discover, so if you're not already connected, you don't even know where to show up.

I think the strongest point here is the compiler-as-feedback-loop argument.

If agents take over more of the middle of the software lifecycle, humans probably move towards the edges: writing better requirements and adding more meaningful tests. This also makes Rust interesiting for a different reason than the usual memory safety pitch. Agents can get a structured feedback based on strict type system, ownership and exhaustive enums in a way that the code either satisfies the contract or it doesn't.

I do wonder if the real distinction is not “simple languages vs complex languages”, but languages that fail early with useful feedback vs languages that allow too much valid-but-wrong code”.