Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Scientists discover new antibiotics using AI

euronews.com

Post reply on HN
Print view View original

61–70 of 100 posts

Re: Scientists discover new antibiotics using AI

#61

Post by graypegg »

rambambram wrote: post #10

What is AI? I've only heard of LLM.

I don't think you should be seemingly in the negatives for this comment. :( It's a totally valid question, and the whole industry has introduced so many new words, acronyms and private definitions. It feels unfair to punish someone for not following them all.

AI is "artificial intelligence". That's a pretty intentionally vague term. It doesn't imply any specific technological implementation, but it does imply the device/software makes decisions autonomously that would otherwise need a human to intervene. There's a million holes in that definition, and that's why the adjective "AI" kind of sucks when used on its own. It's also worth noting it's been used in sci-fi for eons. It's also been used to describe non-playable characters in video games for a long time, which obviously didn't have the sort of "AI" we think of now.

LLMs are neural networks that work with "tokens" from text. So a big statistics-based predictive AI. There are more complex implementations that can consume other kinds of content as well, but "LLM" almost always refers to a neural network that consumes small chunks of text called tokens, and outputs more tokens based on that input. The magic we see with LLMs happens because given enough data, language itself builds up a surprisingly good model of "thinking".

There's some really good explanations in the sibling comments! I'm no expert, and I think someone can probably correct me here if I messed up something there.

Re: Scientists discover new antibiotics using AI

#62

Post by zdragnar »

ac2u wrote: post #6
dvh wrote: post #4

I've been under the impression that we don't have new antibiotics because it is not economically viable. When you invent new weight loss drug you can sell millions of pills to anyone. If you invent new antibiotic it is immediately classified as "reserve antibiotic" and the sale is restricted.

>it is immediately classified as "reserve antibiotic" and the sale is restricted. Is that so medical professionals can break it out as a means of last resort to save a patient and ensure the course is finished? If so that mightn't be a bad idea before we destroy the potential of new antibiotics too by having the general public misuse them render them useless over time.

There's a secondary consideration as well; designing new chemicals is fairly easy, synthesizing them harder, and scaling that to commercial manufacturing very hard.

Oh, and you still have to prove it to be safe and effective through years of clinical trials.

AI is basically still only helping us with the easiest parts, and this particular class is drugs is essentially a limited resource (to avoid force-evolving stains of bacteria that are resistant or immune).

At the end of the day, there's not as much research because there's a finite amount of resources to go around and there are easier / lower hanging fruit.

Re: Scientists discover new antibiotics using AI

#63

Post by CatWChainsaw »

marcusverus wrote: post #30
Waterluvian wrote: post #22

Earlier quoted context omitted.

It requires a socialized solution: to fund it for future needs rather than for profits.

A government which can't regulate effectively can't be trusted with a monopoly. Anyway, how does holding antibiotics in reserve help anything? Why let the bugs evolve resistance to one antibiotic at a time, rather than hitting them them with multiple antibiotics and thus raising the evolutionary hurdle to resistance? Was this strategy designed by the bad guys in a Jacky Chan movie?

I don't know if multiple antibiotics helps prevent resistance necessarily. Resistance usually comes about when the dosage given allows survivors with resistance mutations to become the main population. I mean, I can see where multiple antibiotics and sub-lethal dosages that disrupt function in multiple ways could be enough pressure to break the bacteria, but sufficient dosing is important, and making sure a patient does their regimen properly is entirely the patient's responsibility. In order to prevent the exact same situation playing out with any new antibiotics, we shouldn't allow history to repeat itself.

Plus the normal gut flora would be disrupted by the multiple antibiotics as well, so there would be more work involved in the recovery of the symbiotes and commensals.

Re: Scientists discover new antibiotics using AI

#64

Post by travoc »

wolverine876 wrote: post #33
marcusverus wrote: post #30

Earlier quoted context omitted.

A government which can't regulate effectively can't be trusted with a monopoly. Anyway, how does holding antibiotics in reserve help anything? Why let the bugs evolve resistance to one antibiotic at a time, rather than hitting them them with multiple antibiotics and thus raising the evolutionary hurdle to resistance? Was this strategy designed by the bad guys in a Jacky Chan movie?

Whatever the theory, the government achieves quite a bit. Look at NASA, as one example, not to mention nuclear energy in all its applications, etc.

70-80% of NASA's budget is spent on contracts with commercial companies, research institutions, and other external organizations.

Re: Scientists discover new antibiotics using AI

#65

Post by hoc »

I would have expexted a more generative approach than the one they present.

While this speeds up the screning process quite a lot you might as well feel like someone might just have thrown something out that might actually turn out useful and you can't backcheck without resorting to testing everthing.

When I read the headline I was expected something more along "a network that knows what processes we have and therfor could imagine what compounds we could come up with" combined with that "inverse" network that predicts weaknesses of bacteria. Then those two would be matched and we would have to figure out the missing steps how to actually build the compound coming from that "black box" prediction.

The benefit would be that we would be presented with something that could exist but we hadn't thought of because of our processes and limited resources.

While I value the speedup and cost effect I'd think we could do much better here in exploiting the "creative" and powerful numerical inverse potential of ANNs especially in this area and outsmart those evolutionary skills of bacteria that way.

Re: Scientists discover new antibiotics using AI

#66

Post by WorkerBee28474 »

dvh wrote: post #4

I've been under the impression that we don't have new antibiotics because it is not economically viable. When you invent new weight loss drug you can sell millions of pills to anyone. If you invent new antibiotic it is immediately classified as "reserve antibiotic" and the sale is restricted.

> If you invent new antibiotic it is immediately classified as "reserve antibiotic" and the sale is restricted.

Are there examples of this? I've heard people say this happens, but I've never heard a concrete example.

I googled 'reserve antibiotic' and found Ceftobiprole, and see it's for sale for $40,000/gram, so clearly it's possible to charge money for a 'reserve'.

Re: Scientists discover new antibiotics using AI

#67

Post by stainablesteel »

hoc wrote: post #65

I would have expexted a more generative approach than the one they present. While this speeds up the screning process quite a lot you might as well feel like someone might just have thrown something out that might actually turn out useful and you can't backcheck without resorting to testing everthing. When I read the headline I was expected something more along "a network that knows what processes we have and therfor…

your approach is a lot more complex, but they satisfy a lot of optimal criteria with a simpler approach, which i actually think is more impressive and also a much better target to hit. in fact i think they could have even achieved the same results using a much simpler exploratory method.

a lot of biology, as well as where its been exploited, relies on specific binding, along with this comes the hope that what you test won't inadvertently show off-target effects. Through looking at what's already available for information you can actually use the lack something existing within a human system as a probabilistic edge over the future success of their application. in a regard its like an artistic rendering that makes use of negative space, its an impressive direction.

Re: Scientists discover new antibiotics using AI

#68

Post by wolverine876 »

travoc wrote: post #64
wolverine876 wrote: post #33

Earlier quoted context omitted.

Whatever the theory, the government achieves quite a bit. Look at NASA, as one example, not to mention nuclear energy in all its applications, etc.

70-80% of NASA's budget is spent on contracts with commercial companies, research institutions, and other external organizations.

I'm not sure how that matters. The commercial companies also subcontract.

Re: Scientists discover new antibiotics using AI

#69

Post by fasterik »

UncleOxidant wrote: post #34
dvh wrote: post #4

I've been under the impression that we don't have new antibiotics because it is not economically viable. When you invent new weight loss drug you can sell millions of pills to anyone. If you invent new antibiotic it is immediately classified as "reserve antibiotic" and the sale is restricted.

This is good example of the limits of capitalism.

I think capitalism has plenty of solutions for this. We can allocate government funding for research, pass new laws that create tax incentives, create prize funds to incentivize development of specific drugs, or make advanced market commitments where institutions commit to buying a portion of a certain type of drug after it's produced.

The underlying problem is the lack of political will and cooperation to implement these solutions.

Re: Scientists discover new antibiotics using AI

#70

Post by wolverine876 »

timr wrote: post #60
wolverine876 wrote: post #55

Earlier quoted context omitted.

> Science journalism is abysmal and hyperbolic Do you see the irony? What better describes social media? Personally, I see plenty of good, valuable, useful science journalism; it's got plenty of flaws, like any human endeavor - including social media comments. But attacking journalism is normalized, and thus people know they can do it without being questioned and get some social media likes from it. It would be inter…

> Do you see the irony? What better describes social media? Social media is not one thing. You've got multiple people in this thread who know the subject, in detail, telling you that the headline is wrong. These aren't youtube video comments. > Ironically, that's what the OP is about; that was the point of their research. Yes, I know. That's why I wrote that. The research is about a relatively obscure technical probl…

> Social media is not one thing.

Fair enough, but science journalism is not just one thing.

> You've got multiple people in this thread who know the subject, in detail, telling you that the headline is wrong.

That appears on every HN discussion of anything. It's not a signal of wrongness or rightness, any more than the color of the banner at the top - if it's orange, the OP must be obviously wrong!

Post reply on HN