Everyone now understands the value of AI as a tool for code generation, text translation, summarization, photo generation, and a handful of other simple digital tasks.
But what are the AI use cases that might be possible in the future, yet aren't feasible today for one reason or another?
I'm specifically interested in things that require you to squint out into the future and believe in something that isn't a consensus belief today.
Intelligent, autonomous, self-repairing infrastructure: #1210423
So AI will be used to make sure all the sensors are always lined up correctly?
yeah but you can generalize it:
When you say infrastructure, what other places could this kind of system work?
Can this solve for plumbing, electricity, sewage, etc...?
Yes, I think when you make arbitrary things intelligent and have some kind of body the intelligence can control, it allows for a a lot more solutions. e.g. the location of a solar panel.
My first thought was Tesla's Optimus robots using AI models and vision to perform more complex human labor tasks, but even that is flirting with the "already a consensus belief" line.
I liked #1218570 this morning
Nice, on the surface it looks sort of similar to the Boston Dynamics/Tesla/Figure humanoids, is there a unique angle that this product is taking? Or a new set of use cases that they are aiming to solve?
It's open source, also the hardware. BOM
To choose the Prime Minister of Nepal hahahaha
#1219095
Firefighters, have them control strong humanoid robots with 100+ censors, so they (the firefighters) can do their job more safely.
Good idea, could also probably work for police officers
Maybe it's because I don't see AI as a new thing in a material sense, because if you distill it down and it's just another algorithm... but I think eventually we're realize what we're actually using it for already... an interface to other algorithms.
We see it in agents and chatbots now, things like MCP, where natural language calls upon external API's in tools which then have their own algorithm or process.
This ability to chain functions from disparate sources makes it a new interface type for applications that are just a representation of many functions.
There's a lot of things we think of in terms of UI now that will probably just become MCP-adjacent tools in a language interface vs. a visual one, image gen vs. Photoshop for example. It's already consolidating time away from the browser (RIP keyword searches and stackoverflow), and voice based bots can obviate screens altogether in many cases.
We're probably less than a generation away from kids not knowing wtf a keyboard and mouse is. A pocket projector that can leverage your eye/hand/body movements, or glasses, will change how much we think about interfacing as the iphone did.
Interesting thought experiment
I would say that given how certain countries around the world have legalised euthanasia, there might come a point in time in which AI capabilities are incorporated into the process to ascertain whether one’s application process for euthanasia should/should not be approved
AI analysis is already used in many expert systems in high end hospitals. However, these aren't of the LLM kind.
Albania with a new AI minister to fight corruption lol
Automated news scanning to highlight events that are a threat to the status quo.
Like https://www.ranenetwork.com/platform/products/threat-intelligence?
I used an earlier version of it to assess safety for my teams and I when I was working globally.
I think the future of AI is already here in our space exploration, healthcare, supply chain. And AI is replacing humans in many areas of digital tasks and physical work too. It's best time to leverage and utilities the full potential of AI in an ethical and responsible way for our future generations.