raychis

First thing is first, this is really cool. This feels like the right way to frame LLM-assisted engineering. AI can generate a shocking amount of code, but the actual value is in the review discipline, and tests around it. The browser Kubernetes angle is cool, but what I find more interesting is the workflow, and especially testing behaviour against k8s instead of just trusting “looks right.” I do wonder how many teams are already doing this level of verification for AI-written code. It might be the direction everyone goes in over the next few years.

Yeah, I 100% agree with you. I was very reluctant to pick up AI tools for coding for a long time. But now I have, I've gone from thinking the code was the most important bit to realising the thinking behind the code is the actual real important bit. I am not sure AI tools will ever be able to replace that element.

This is a good thesis but it does lean a bit too hard on a vague 80/20 metaphor. It kind of romanticises old-school engineering struggles while downplaying how much of past learning was just wrestling with crappy tooling or poor docs. Things are much better now, I wouldn't want to go back. The stronger argument would not just be the old way is better, but that we need a way to preserve judgment that used to be developed through the struggle.

The bar for scientific discoveries is generally incredibly high and rightly so. In particle physics for example for a discovery to be confirmed it must be independently verified by a second experiment and both the discovering experiment and confirming experiment must be within five standard deviations (or a roughly 1 in 3.5 million) chance of not being a discovery. Its not a perfect comparison because astrobiology is messier but it is useful to see how high the bar is.

For a discovery of the magnitude of life outside our planet the scientific community would probably need something like: a real and well-characterised signal, multiple independent lines of evidence, strong geological context, replication by independent teams, and a serious failure of plausible alternate explanations. Until then, “potential biosignature” is probably the right phrase.

To be clear, I think like you that there probably was life at some point on Mars. I am pretty convinced. But, you and I being pretty convinced does not come close to the threshold of a scientific discovery.

Hurrah for more terrible authoritarian policies out of the US government. Really great for the US, really great for humanity.

Not just America, the rest of the world. There are absolutely people who will get left behind. But, I still do think it remains to be seen if the current AI boom will really sustain itself or fizzle out with only. It is worrying none the less that people and organisations could be left behind, if that does happen it will have massive societal impact.

This is an odd comment. Any scientist or scientific organisation would love to be the first to discover life on another planet. It would catapult that organisation and individuals involved to legendary status with their discoveries being counted amongst the greatest of humankind. It would be an epoch defining moment. Funding for their work and personal riches would pour in. There would be movies made about it. Their names would be remembered thousands of years into the future.

To imply there would be a conspiracy to cover up such discoveries because you think the opposite would happen is such an odd way to think about things.

The article is good but the title is a bit too slippery a statement in my opinion. The article is saying more evidence is consistent with possible ancient life on Mars. In astrobiology the massive problem is that geology can imitate biology. The presences of minerals formed by microbes on Earth does not prove microbes are involved in their production on Mars, it is a big jump to make.

This has been a massive frustration for me at work. The amount of emails I get that are drafted by co-pilot or other AI models. They are so obvious to me and they are so generic and clearly missing context that any human paying attention would know. They end up being nonsensical. It is exhausting.

This is really great. I especially like the explanation of how claims are built from primary sources, and the tool clearly aims to show cause-and-effect relationships rather than just present conclusions.

My only suggestion would be to tone down rhetorical phrases like “science dictates economics.” The tool is strong enough to stand on its own, and I think presenting the data more neutrally would make it even more persuasive.

Yes please! Thank you to whoever created this.

Wow, always suspected this was going on. But, to read it like that in black and white it is really quite shocking.

Data harvesting has got completely out of control. If only we had an organisation with the power to stop it by making something like laws...

The UK government's turn towards authoritarianism on these sorts of things is extremely worrisome! Invest in a VPN to keep yourself safe.

On the fence with this. Social media has been show to be extremely damaging to young developing brains, something needs to be done about that. But, the privacy breaches and governmental surveillance this could potentially entail is very worrying.

The web from the 90s was so much more fun. I have a vivid memory of a website from some guy dedicated to Liverpool FC and his favourite dinosaurs, you would never get that kind of random fun any more.