One of the prevalent doomerist ideas about AI is that big tech will control it all in the future. Yet reality is behaving totally differently. Open-source AI seems only a few months behind big tech.
I wonder when this is going to seriously affect world oil demand? People used to think "Peak Oil" would be when supply was constrained, it turns out it will be when demand is constrained.
After years of being "almost there" 2023 seems the year self-driving robot vehicles have finally come of age. In several cities in the US & China, people can hail self-driving taxis within city limits. It surprises me that fixed route buses, like the model talked about here, aren't taking off faster. In many ways they are simpler than self-driving taxis, needing only level 4 self-driving. They are also an incredibly obvious solution to help reduce fossil fuels. A self-driving bus network with buses stopping every 5 minutes that served a city's busiest 100 locations would make many people ditch car journeys.
There's widespread awareness in 2023 that AI has started advancing extremely rapidly, but there's less awareness of many of the second-order effects resulting from that. One of those is that AI advances will quickly feed into robotics advances. Robots after all are pure AI embodied into our 3D physical world.
Another thing I don't think people appreciate is that robots of the future may be cheap and ubiquitous. We tend to think of them as humanoid, relatively rare, and somehow "special" because of their near-humanness. Data in Star Trek is a well-known embodiment of this idea.
But what if future robotics is dominated by hundreds of millions of small animal-sized robots, and maybe billions or hundreds of billions of insect-sized robots?
AI is making content automation at vast scales so easy it seems what is just a trickle now will inevitably become a deluge. What's the solution? Perhaps using AI to moderate the AI content? I think there's a market for someone to make personal AI moderator software that filters out all these people and their garbage from wherever you are on the internet. I'd sign up for it.
This might sound frivolous at first, but the more you think about the less that it seems so. Prada are famed for their minimalist aesthetic. If they can successfully bring that to spacesuit design it will be a bonus. But they can add more. They have expertise in materials and garment manufacturing that space companies won't have. It will be interesting to see what comes out of this collaboration. It might become a "classic" space suit design, used into the 2030s and 40s.
This is a great idea on paper, but I'm curious to know how they can make it a reality. Realistically, there are only a few companies in the world who can pull off manufacturing this. They have two former senior Airbus people on their advisory board.
I'm curious to see where they go with this. I'm guessing making it a reality will cost a LOT of money. For context - Airbus Group research and development expenses for the twelve months ending June 30, 2023 were $3.422 billion.
Is the play here to hope they get bought out by the likes of Airbus or Boeing? I don't know, but they do seem to have spotted a gap in the market and the idea seems sound.
A common dystopian narrative around AI is that Big Tech will own all the gains, yet reality doesn't match up to that. What we're seeing instead, is that freely available open source AI isn't far behind the leading contenders.
What will it mean for the future? Who knows. But I feel reassured when it comes to AI power is being more decentralized.
Glad to see this but I'm more interested in Level 4 buses for public transport. These are starting to pop up around the world, and mean driverless buses can follow fixed routes without drivers. When are we going to see cities served by thousands of these vehicles? That would revolutionize public transport.
Google's search page has got noticeably worse in recent years, for a long list of reasons - here's another indication it's going to get even worse. I find myself using Duckduckgo more and more - it has its problems, but they are not as bad.
I think its real potential (as yet unexplored) is in disintermediation.
Two parties, who do not know each other, and want to conduct business rely on so many intermediaries who provide services, and crucially, trust - who all take their cut. Lawyers, bankers, etc
Blockchain could not only automate the processes, but also be designed to replicate trust.
Also, that potential isn't just about efficiency - its about power, and taking it away from industries and institutions - banking, etc
There are smaller models that can run on most laptops.
https://www.maginative.com/article/stability-ai-releases-stable-lm-3b-a-small-high-performance-language-model-for-smart-devices/
In benchmarks this looks like it is not far off Chat-GPT 3.5.