Tag Archives: artificial intelligence

October 2024 in Review

Leave a reply

Only half way through November – here is an “October in Review” post.

Most frightening and/or depressing story: When it comes to the #1 climate change impact on ordinary people, it’s the food stupid. (Dear reader, I’m not calling you stupid, and I don’t consider myself stupid, but somehow we individually intelligent humans are all managing to be stupid together.) This is the shit that is probably going to hit the fan first while we are shouting stupid slogans like “drill baby drill” (okay, if you are cheering when you hear a politician shout that you might not be stupid, but you are at least uninformed.)

Most hopeful story: AI, at least in theory, should be able to help us manage physical assets like buildings and infrastructure more efficiently. Humans still need to have some up-front vision of what we would like our infrastructure systems to look like in the long term, but then AI should be able to help us make optimal repair-replace-upgrade-abandon decisions that nudge the system toward the vision over time as individual components wear out.

Most interesting story, that was not particularly frightening or hopeful, or perhaps was a mixture of both: Some explanations proposed for the very high cost of building infrastructure in the U.S. are (1) lack of competition in the construction industry and (2) political fragmentation leading to many relatively small agencies doing many relatively small projects. Some logical solutions then are to encourage the formation of more firms in the U.S., allow foreign firms and foreign workers to compete (hardly consistent with the current political climate!), and consolidate projects into a smaller number of much larger ones where economies of scale can be realized. There is some tension though between scale and competition, because the larger and more complex a project gets, the fewer bidders it will tend to attract who are willing to take the risk.

September 2024 in Review

Leave a reply

I was sitting down to do my “October in Review” post and realized I never got around to September. So better late than never. I’m writing this on November 9, 2024 after the U.S. election but I’ll try to give U.S. politics a rest in this post (update: I almost succeeded although I couldn’t resist an interesting point about the U.S. Constitution).

Most frightening and/or depressing story: There is nothing on Earth more frightening than nuclear weapons. China has scrapped its “minimal deterrent” nuclear doctrine in favor of massively scaling up their arsenal to compete with the also ramping up U.S. and Russian arsenals. They do still have an official “no first strike” policy. The U.S. by contrast has an arrogant foreign policy.

Most hopeful story: AI should be able to improve traffic management in cities, although early ideas on this front are not very creative.

Most interesting story, that was not particularly frightening or hopeful, or perhaps was a mixture of both: Countries around the world update their constitutions about every 20 years on average. They have organized, legal processes for doing this spelled out in the constitutions themselves. The U.S. constitution is considered the world’s most difficult constitution to update and modernize.

AI and asset management

Leave a reply

This article is about AI and predictive building maintenance. It also reads like an IBM corporate press release, but nonetheless it sparks some interesting thoughts. Recently I was at a conference where a friend of mine was on stage and as asked what technologies would be most important for the future of public infrastructure (water infrastructure, in the case of this particular conference.) AI and asset management came to my mind, and I willed my friend to also think of this. Alas, he did not. Now, if I had been up there would I have been able to articulate my thoughts clearly on the spot? Probably not, but with the benefit of a few minutes to think here is what I fantasize I might have said.

Basically, AI should be pretty good at asset management. Given good data on assets and their ages, they should be able to identifying assets (we’re talking physical assets here, like pipes or electrical equipment, or even green infrastructure like street trees) that are nearing the end of their service life and likely to fail in the engineering sense of no longer serving their intended purpose efficiently. Or, somewhat obviously, when things really have failed AI can help get that information to the attention of whoever can actually do something about it. Well, I still think humans have to do the up-front planning and have some vision for what they would like the infrastructure system to look like 20, 30, 50 years down the line. But then, AI should really be able to help with those repair-replace-upgrade-abandon decisions, so that as things wear out the system is slowly nudged in the direction of that long-term vision, all while minimizing life cycle cost and balancing whatever other objectives the owners or stakeholders might have. This all looks good on paper and is messy to do with a mish-mash of real-world governments and institutions and companies, but having the vision is a start.

AI-controlled stop lights

Leave a reply

Boston and other cities have piloted tested AI-controlled stop lights and found that they can reduce “stop and go traffic”. This seems encouraging to me, but not very imaginative and I hope this is not the end of the story. Stop lights are such an old technology, and it seems to me that with modern LED lights and screens we should be able to do much better. Each traffic lane, including lanes dedicated to light and un-motorized vehicles, need their own signals. Let’s get rid of the colored circles and make every single traffic light a series of arrows, so that we can control who is allowed to go forward entirely separately from who is allowed to make a turn from each lane. Pedestrians also need their own signals, and the heavy/highway vehicles, light/unmotorized vehicles, and pedestrians must never, ever have signals that put them in the same space at the same time. I won’t buy the idea that this would be “too expensive” – I happen to be traveling in a middle income country at the moment and I see a lot more arrows and countdown timers on traffic lights compared to what we have in the U.S. (although the jurisdiction I am in is no traffic safety utopia for sure.) If this sounds like it would be too inefficient with today’s system, this is where AI should come in and make it efficient and safe, at each individual intersection and for the system as a whole.

Another level of science fiction would dispense with the lights and screens, and embed them in our vehicle windshields, augmented reality glasses, headphones, etc. Vehicles that are entirely computer controlled, of course, can just get their signals from cellular or wireless networks. We are not there yet at least when it comes to widespread access/adoption of these technologies, but the technologies themselves either exist or are on the horizon.

July 2024 in Review

Leave a reply

Most frightening and/or depressing story: Joe Biden’s depressing decline in the international spotlight, and our failed political system that could let such a thing happen. Not much more I can say about it that has not been said. The “election trifecta” – non-partisan, single ballot primaries; ranked-choice general elections; and non-partisan redistricting – is one promising proposal for improving this system.

Most hopeful story: A universal flu vaccine may be close, the same technology might work for other diseases like Covid, HIV, and tuberculosis.

Most interesting story, that was not particularly frightening or hopeful, or perhaps was a mixture of both: Maybe we could replace congress with AI agents working tirelessly on behalf of us voters. Or maybe we could just have AI agents tirelessly paying attention to what the humans we have elected are doing, and communicating in both directions.

AI Biden

Leave a reply

I can’t tell if this post is serious or not, and this is the mark of bad satire.

Despite an ambitious and widely praised first term in office, he is currently trailing in polls to a man who incited an insurrection and was recently convicted on 34 felony counts. Something needs to change, and much to the chagrin of West Wing fanatics in the beltway, it won’t be the Democrats’ 2024 nominee. Modern technology offers a clear solution. AI can be used to polish how the president comes across, allowing voters to focus on his substance. How many times have we heard voters and pundits alike gripe that “Biden would be the perfect candidate if he were just 10 years younger?” With modern technology, this exact deliverable is possible.

huffpost.com

I sincerely hope this is intended as irony. I have thought however about an AI-based experiment in direct democracy. In this concept, an AI agent would represent me, the individual citizen. It could spend time patiently interviewing me about my views and opinions, and then it could go to Congress and negotiate and vote on legislation with the AI agents of the 328 million other Americans. If I really have time, I could take over control of the agent at any time I want, then hand the keys back over to it any time I want.

A watered down version of this could be AI constantly talking to an elected representatives constituents about their views on various issues and the content of proposed legislation, patiently explaining to the elected representative how his or her actual constituents would like him or her to vote (the pronoun thing is exhausting, and yes I know I have only scratched the surface of potential pronouns), and patiently explaining to the constituents how it all turned out. Imagine if a politician made it a campaign pledge to always vote according to the wishes of a majority of their constituents (as interpreted through the AI agents) no matter what.

I thought of this a long time ago, with the idea that it could be done through polls or through an app, but the natural language AIs could make this much more practical and achievable by just chatting with humans for a few minutes each day and providing constant feedback.

Now, of course the problem with direct democracy is always that 51% of the people might want to exterminate or enslave the other 49%, which is one way to achieve consensus but not what we are looking for. So you obviously have to couple this with protections for the rights of minorities and human rights in general.

how I’m using AI

Leave a reply

AI has definitely improved my personal productivity when it comes to computer programming. I haven’t been successful asking it to write whole programs for me, but it has been fantastic for solving syntax problems in minutes that might otherwise take me hours to figure out. For example, I have data in xyz format and I need it in zyx format, please give me some example code that works. Or, I need to pass an argument to a function and does it need to be in quotes, parentheses, enclosed in ancient hieroglyphics or some random combination of these? In the past, I always started with a Google search on these questions, looking first for a blog post with examples, and failing that for a Stack Overflow post. At some point, I started using ChatGPT when those two options failed. Then I figured out I have access to a version of CoPilot through my employer and any data or code I supply is not going to be automatically broadcast to the world, so I have gradually been shifting to that. I just learned that CoPilot is really a version of ChatGPT. (The article I linked to mentions some other AIs I had not heard of yet, such as “Claude”.)

Then at some point, I started going to AI after blog posts but before Stack Overflow. This is about where I am now. For one thing, AI tends to listen to my question, understand what I am looking for and give me a relevant answer much more often than Stack Overflow. For another, it is much more polite than the dick heads and whining weenies on Stack Overflow. You know who you are. Thank you for your free service in the past, and if you want me to continue coming to you, you may want to at least learn some manners. You could start by asking an AI to analyze your posts and suggest ways to not be such a dick head.

I am not using AI for writing, because for me writing and thinking are two halves of the same coin, and I can’t farm out the thinking. The one exception to this is thank you notes and other social niceties – I have no interest in burning my limited intellectual capacity on learning how to write these, so I am very happy to have AI do it. I tried asking CoPilot to find me promo codes for a few stores, but none of them worked. I suspect the companies are paying for the same AI I am using for free, so it is probably snitching to them so I can’t get a deal.

GPT-4 and the Turing test

Leave a reply

I don’t know if the original Turing test was based on just one human participant, or if more than one was used, how those humans were chosen.

Scientists decided to replicate this test by asking 500 people to speak with four respondents, including a human and the 1960s-era AI program ELIZA as well as both GPT-3.5 and GPT-4, the AI that powers ChatGPT. The conversations lasted five minutes — after which participants had to say whether they believed they were talking to a human or an AI. In the study, published May 9 to the pre-print arXiv server, the scientists found that participants judged GPT-4 to be human 54% of the time… GPT-3.5 scored 50% while the human participant scored 67%.

livescience.com

Should the criteria to pass be that 51% of a large random sample of humans could not correctly identify computer vs. human? How bad would the results have to be for the control (identifying the human as human) before we would conclude that the Turing test no longer makes sense?

It’s interesting that the Turing test is presented as a test of intelligence, but many of the things that apparently make computer conversationalists convincingly human are in fact cognitive biases, logical errors, and the appearance of emotionally-influenced decision making. These might be things you would look for if you wanted a computer to be a friend, but they are not things I would like for if I wanted a computer to counter my less rational human impulses and help me make more rational decisions.

AI and protein research

Leave a reply

Here is a story in MIT News about AI doing experiments on proteins, with drug development and gene therapy implications. This seems like the clearest application of AI at the moment – anything where there is a formula to be figured out and a large number of combinations to be tried. I can definitely see this accelerating scientific and technological progress, although the efficiency to me seems to be more in the “automation” part than the “intelligence” part.

trade “fragmentation” vs. AI?

Leave a reply

One interesting thing in the IMF report I mentioned recently forecasting a significant productivity slowdown: the positive effects of AI on productivity and the negative effects of inefficient trade policy were shown offsetting each other. Meanwhile, Eric Posner is concerned that humans will have psychological difficulties leading lives of leisure after the AI-driven productivity revolution, and after our political system correctly decides to redistribute the resulting wealth to everyone. I know, this could be a medium-term pain, long-term gain sort of thing. But how do we know the long term will come? And this kind of thinking clearly ignores the existential threats like climate change and biological weapons, unless you assume the AI productivity revolution will dispatch those threats without creating new ones.