MLMs are crazy useful in processing massive amounts of data for a bunch of scientific applications, and AI in general has uses in manufacturing and infrastructure, but the mass marketed LLM slop is not that and absolutely not necessary (or even particularly useful) for an individual.
Yea I tend to use “AI” as a term for generative AI, which I should stop cause of how broad it is. I definitely see the usefulness of machine learning models for research purposes. I haven’t heard of how it’s used in manufacturing and infrastructure though, interested in looking more into that.
Its most easy to understand application in manufacturing is in creating more efficient structures, as a ratio of strength to material used. It’s called “generative design”, which is kind of an alternative to another interesting technique that doesn’t typically use AI called “topology optimization”, if you want to look at both of those. Unsurprisingly, they both end up looking very “organic”.
Kumikommunism already talked about manufacturing, but they do (/could be used to do) a lot of optimisation in infrastructure - stuff like ai-controlled traffic lights to relieve traffic flow, detecting vibrations in buildings or bridges to predict where preventative maintainance will be needed, energy demand prediction and distribution - stuff where the data sets are naturally massive or require extreme precision are perfect for AI. Then there’s the ai-powered “dark factories” that just need some attention from an engineer now and then, operating almost completely autonomously.
MLMs are crazy useful in processing massive amounts of data for a bunch of scientific applications, and AI in general has uses in manufacturing and infrastructure, but the mass marketed LLM slop is not that and absolutely not necessary (or even particularly useful) for an individual.
Yea I tend to use “AI” as a term for generative AI, which I should stop cause of how broad it is. I definitely see the usefulness of machine learning models for research purposes. I haven’t heard of how it’s used in manufacturing and infrastructure though, interested in looking more into that.
Its most easy to understand application in manufacturing is in creating more efficient structures, as a ratio of strength to material used. It’s called “generative design”, which is kind of an alternative to another interesting technique that doesn’t typically use AI called “topology optimization”, if you want to look at both of those. Unsurprisingly, they both end up looking very “organic”.
This is the first article I found: https://parametric-architecture.com/nasa-uses-ai-to-design-mission-hardware-that-looks-somewhat-alien-and-weird/
That’s actually really cool
Kumikommunism already talked about manufacturing, but they do (/could be used to do) a lot of optimisation in infrastructure - stuff like ai-controlled traffic lights to relieve traffic flow, detecting vibrations in buildings or bridges to predict where preventative maintainance will be needed, energy demand prediction and distribution - stuff where the data sets are naturally massive or require extreme precision are perfect for AI. Then there’s the ai-powered “dark factories” that just need some attention from an engineer now and then, operating almost completely autonomously.