AI “coscientist”

The idea of computers and robots greatly accelerating the rate of progress in chemical and drug research is not science fiction.

Autonomous chemical research with large language models

Transformer-based large language models are making significant strides in various fields, such as natural language processing1,2,3,4,5, biology6,7, chemistry8,9,10 and computer programming11,12. Here, we show the development and capabilities of Coscientist, an artificial intelligence system driven by GPT-4 that autonomously designs, plans and performs complex experiments by incorporating large language models empowered by tools such as internet and documentation search, code execution and experimental automation. Coscientist showcases its potential for accelerating research across six diverse tasks, including the successful reaction optimization of palladium-catalysed cross-couplings, while exhibiting advanced capabilities for (semi-)autonomous experimental design and execution. Our findings demonstrate the versatility, efficacy and explainability of artificial intelligence systems like Coscientist in advancing research.

Nature

It seems to me that the speed limit here is not anything imposed by the computers and robots, but your ability to measure progress and give the computers and robots feedback. With chemicals, you could tell the robots to find a combination of compounds that will do XYZ, where XYZ is something you can measure like an amount of energy or a color. With drugs, your issue could be how to test the results to see if they are working. If you test them on a computer model, your ability to measure depends on how good the computer model is. Let’s say you wanted to breed a super-intelligent mouse. There should be ways to measure the intelligence of a mouse. So you could take 100 mice test them all, find the two smartest and create a new batch of 100 embryos from the smartest male and female (or maybe at some point gender is no longer a limitation?). Now you have to wait for those 100 embryos to grow up to the point you can repeat the process. The limiting step here would be how long it takes the mice to develop to the point they can be tested. If they could somehow be tested at the embyro stage, maybe you could create a thousand generations of mouse directed mouse evolution in a matter of hours or days? Well, then, you can let the super-intelligent mice design the next round of robots.

Bill Gates on 2023

Bill Gates’s year-end retrospective is kind of rambling but here are a few points I pulled out:

  1. Lots more vaccines were administered to children in developing countries using new technologies and new delivery methods. This has made a big difference in child mortality, and that is always a happy thing. He doesn’t really go into details on the new technologies, but I am imagining things like nasal sprays rather than needles, and vaccines that don’t require refrigeration or not as much. And sometimes we just figure out how to make familiar things but make them much cheaper, and this can make a huge difference. Which would illustrate that important technologies don’t have to seem extremely complicated and high-tech to have a big impact.
  2. On the AI front, he says it will accelerate drug development, including solutions for antibiotic resistance. I don’t doubt this, although I suspect the hype has gotten a bit ahead of the rollout. So I would look for this over the next half-decade or so rather than expecting it to burst on the scene in 2024. Bill actually predicts “18–24 months away from significant levels of AI use by the general population”.
  3. He talks about AI tutors for students. I don’t want to be a Luddite, but I am concerned this will just mean less teachers per student, which will be bad.
  4. Maybe AI can just get our medical records under control. This would be nice. Transparent, common protocols for how medical records should be formatted, stored, and shared could also do this though. I can someday hope robots will constantly clean up and organize my messy house as I just throw my things everywhere, or I could organize my house (which would take a big effort once) and keep it that way (which would take small, disciplined daily efforts).
  5. Gut microbiome-based medicine. Sounds good, I guess. Then again, whenever we try to replace nutritious whole foods with highly manufactured alternatives (vitamin pills, baby formula) we tend to decide later that we should have stuck with the whole foods.
  6. “a major shift toward overall acceptance of nuclear” power. Well, it’s been pretty obvious to me for a long time that this had to happen, but maybe the world is catching up. Nuclear could certainly have been the bridge fuel to renewables if we had fully adopted it decades ago. The question now is whether, given its incredibly long time frames to get up and running and the fact that any technology is obsolete by the time it is up and running, and the current pace of renewables, it still makes sense. I definitely think we should put some eggs in this basket though.
  7. He mentions the fusion breakthrough at Lawrence Livermore about a year ago. It’s been a year and we haven’t heard much more – is the time to refine and rollout that technology going to be measured in years, decades, or never?
  8. He talks about the need for more investment in electric grids and transmission lines. Yes, this is unsexy but really needs to happen. Will it?

2023: a year of war?

Richard Haas says 2023 was characterized by the wars in Ukraine and Gaza, and also by the world starting to wake up to the climate change threat (but falling far short on its response.) His bright spots are that the U.S.-China relationship did not devolve into a hot war, and Japan and South Korea are getting along a bit better. To me, it is unforgivable that world leaders are leading us towards war when we have such dire challenges staring us in the face including food security, loss of major coastal cities, nuclear proliferation, and pandemic threats both natural and manmade. Climate change is also fueling a vicious cycle of migration, anti-migration political movements, and geopolitical instability, which in turn makes it even harder to focus on solving problems together.

stuff I’ve read in 2023

This is just a grab bag. I’ve read two reasonably entertaining novels set in near futures where climate change is ravaging the world. Neal Stephenson’s Termination Shock was the more entertaining of the two. Stephenson is a good storyteller and his books are easy to read. But obviously, read Snow Crash first if you never have.

I’m about half way through Ministry of the Future by Kim Stanley Robinson. Like all Robinson’s books I have read so far, this one is less entertaining but he does a good job of world building. I feel like he is trying too hard to educate me though – sometimes if I want to be educated I will go read a non-fiction book, and if I want to be entertained I will turn to fiction. Books that lie somewhere in between can be irritating.

I’ve enjoyed two mystery/action series this year. The Jack Reacher series is great escape fiction about your basic middle aged white male hero type who is good with guns and fighting, and fighting with guns. I hate violence in real life but it is hard to write entertaining stories with zero violence, so there. I don’t think I ever want to see the movie because I enjoy the book character too much. I’ve also been enjoying the Bernie Gunther series. This one is is bit dark and morally complex as it deals with a detective/policeman hero type who happens to be an involuntary reluctant Nazi. But seriously, it is good and recommended.

Speaking of dark and morally complex, I read a couple books by Octavia Butler – Kindred and Parable of the Sower. She subjects her characters to niceties like murder, rape, and slavery, and I guess you get to find out what they are made of when they respond to these situations. She is a good character developer and storyteller though and worth a read, as long as you are not already depressed going into it.

I read Dan Simmons’ Ilium and Olympos series, which is about Greek Gods, transhumans, Greek Gods who may be transhumans, robots, Shakespeare, Proust, and robots who like Shakespeare and Proust. This is pretty crazy stuff and you have to really like Dan Simmons to like it. I am liking but not loving it. I guess I would read the Hyperion series first, if you have not read any Dan Simmons.

Actually what prompted this post was a post by Charlie Stross (contains spoilers) about his Laundry Files and New Management Series. I gobbled up the new one that came out this year, Season of Skulls, because I gobble up all his stuff as soon as I can. It’s exciting to hear Charlie mention in this post that he has plans to wrap up the Laundry Files series with two more books “because his publisher insists”. (Could George R.R. Martin sign on with this particular publisher?) If you haven’t read any of these, I would go all the way back to the beginning of the Laundry Files, and enjoy!

most popular R books of 2023

Here is something useful (to me, personally, and maybe too others), and thankfully not too pessimistic or morally fraught.

A Crash Course in Geographic Information Systems (GIS) using R – yes, please! We must end the tyranny of the monopolistic Environmental Systems “Research Institute”. Okay, they make some nice products, but just admit you are a rapacious for-profit corporation, please!

A ggplot2 Tutorial for Beautiful Plotting in R – Who doesn’t need to improve their data visualization and communication game?

“if there’s one thing we don’t want here in South America it’s war”

This seems like a sensible quote from President Lula of Brazil. But countries that threaten to or actually nationalize lucrative industries controlled by U.S. based companies (Cuba, Iran, and Iraq come to mind) have a tendency to get invaded by the U.S. I know next to nothing about the politics of South America, but I do know the U.S. establishment has been itching for a fight with Venezuela’s Nicolas Maduro for quite a while, so this seems like it would be a huge self-inflicted wound for a country already going through a lot of turmoil. And the world clearly does not need another war on another continent right now. Hopefully cooler heads will prevail.

sexiest people sorted by pronoun, 2023 edition

The 2023 sexiest person of the year, somewhat obviously, is Taylor Swift, who is also the Time Person of the Year. I can’t argue with that too much, although I think it tells us something whether the person of the year comes from pop culture rather than than the serious worlds of, say, politics or science. At the moment, I think it tells us that we are burnt out on serious things and we want to stick our heads in the sand and ignore them.

Who would I pick, if I were going to pick someone from the serious world? That’s a tough one. I almost want to tip my hat to Jimmy Carter because we really need serious voices for peace in the world right now. But then again, Nobel Prize winners might not want to be greedy. I may have to keep thinking.

Meanwhile, it is mildly ironic that “People” magazine still names a sexiest “man”. I don’t really have a problem with this – People magazine doesn’t pretend to be about anything other than pop culture. And I still see men here and there every day despite the various endocrine disrupting compounds in our water and consumer products. This year’s sexist middle aged Caucasian actor I haven’t heard of is Patrick Dempsey. Congratulations, Patrick. I am sure you are awesome, it is just that I am not paying attention to your particular corner of pop culture. I note that you have a few gray hairs, so maybe my chances of being the sexiest man alive at some point in the future are not completely dead.

Q (the AI)

“Q star” is very badly named, in my view, given the “Q anon” craze it has absolutely nothing to do with. Then again, the idea of an AI building an online cult with human followers does not seem all that far fetched.

Anyway, Gizmodo has an interesting article. Gizmodo does not restrict itself to traditional journalistic practices, such as articles free of profanity.

Some have speculated that the program might (because of its name) have something to do with Q-learning, a form of machine learning. So, yeah, what is Q-learning, and how might it apply to OpenAI’s secretive program? …

Finally, there’s reinforced learning, or RL, which is a category of ML that incentivizes an AI program to achieve a goal within a specific environment. Q-learning is a subcategory of reinforced learning. In RL, researchers treat AI agents sort of like a dog that they’re trying to train. Programs are “rewarded” if they take certain actions to affect certain outcomes and are penalized if they take others. In this way, the program is effectively “trained” to seek the most optimized outcome in a given situation. In Q-learning, the agent apparently works through trial and error to find the best way to go about achieving a goal it’s been programmed to pursue.

What does this all have to do with OpenAI’s supposed “math” breakthrough? One could speculate that the program that managed (allegedly) to do simple math operations may have arrived at that ability via some form of Q-related RL. All of this said, many experts are somewhat skeptical as to whether AI programs can actually do math problems yet. Others seem to think that, even if an AI could accomplish such goals, it wouldn’t necessarily translate to broader AGI breakthroughs.

Gizmodo

My sense is that AI breakthroughs are certainly happening. At the same time, I suspect the commercial hype has gotten ahead of the technology, just like it did for every previous technology from self-driving cars to virtual reality to augmented reality. Every one of these technologies reached a fever pitch where companies were racing to roll out products to consumers ahead of competitors. Because they rush, the consumer applications don’t quite live up to the hype, the hype bubble bursts, and then the technology seems to disappear for a few years. Of course, it doesn’t disappear at all, but rather disappears from headlines and advertisements for a while. Behind the scenes, it continues to progress and then slowly seeps back into our lives. As the real commercial applications arrive and take over our daily lives, we tend to shrug.

So I would keep an eye out on the street for the technologies whose hype bubbles burst a handful of years ago, and I would expect the current AI hype to follow a similar trend. Look for the true AI takeover in the late 2020s (if I remember correctly, close to when when Ray Kurzweil predicted 30-odd years ago???)

The Sierra Madre

This is a weird story. In the South China Sea, there is a Philippines ship that ran aground on a shoal in 1999. Sailors from that country have occupied the ship ever since, and are regularly resupplied while being bombarded by the Chinese navy using things other than guns, like water cannons and lasers.

That month, the Sierra Madre ran aground at Second Thomas Shoal, a small reef in what was then disputed territory, about 120 miles off the coast of Palawan island. A second ship did the same at another shoal later that year. Beijing suspected that Manila was using the beached ships to create outposts.

Philippine officials initially played coy, saying that they meant to repair the Sierra Madre but were having trouble finding the materials, while the other ship was eventually towed away. Yet, more than two decades later, the Sierra Madre remains grounded, a rusted dieselpunk monolith interrupting an otherwise pristine swath of tropical waters. A small group of sailors crews it; they pick their way through its slightly listing steel skeleton as they monitor the area for incursions. Their rotations generally last two months but can stretch up to five. Carlos referred to these tours as a “test of sanity…”

Beijing blatantly ignores this ruling. When the Philippines delivers supplies for the sailors on board the Sierra Madre via small boats escorted by coast-guard ships, Chinese ships attempt to block them. In early August, the Chinese coast guard used water cannons to prevent Philippine boats from reaching the outpost. A second attempt later that month was successful, as was one in September, when a U.S. reconnaissance aircraft flew overhead.

Atlantic

The U.S., of course, feels the need to get involved in all this. Not by being a voice of reason, but by ramping up tensions and threats of violence.

new CDC Covid dashboard

CDC has released a new wastewater monitoring dashboard, according to Forbes, which oddly does not link to the new dashboard or even tell us the name of it so it would be easy to find. Wastewater monitoring to me seems like a very useful tool in the toolbox to monitor public health and diseases of concern, whether common or exotic, well known or emerging.