Comment by doku | Hacker News Reader

Comment by doku | Hacker News Reader

90% confidence level, Hottest weather in recorded history. Extreme weather in many places not seen before. No one is prepared for this commonplace outlier, Globally common, locally outlier. Huge insurance payout but most not fullfilled.

90% CL: Top search engine results completely inundated by LLM text, we none the wiser. SEO solved. Search engine switches to new algorithm based on LLM.

90% CL: AI with tool use that has access to physics simulator and CAD software will automatically advances efficent engine design.

70% CL: Million dollar prompt run on million dollar math prize. Millennium Prize Problems

40% CL: AI sucessfully held back invasive species encrochment with lessons from AI war machines

30% CL: Cheap AI nose sniffs out many common diseases. Automatic data analysis of Gas Chromatography and spectroscopy readings on patients' VOC(volatile organic compounds) produces fast and cheap diagnosis.

10% CL: Not next year, Cost of renewable electricity in remote area becomes cheap enough to pull oxygen from air or electrosis from water to be sold as commodity. Or other processes to package solar energy for trade.

Renewable energy has been the cheaper option in remote areas for a while now, solar demand in Pakistan has skyrocted recently.

Solar powered water pumps and water purification has been a commodity and charity item since 2023 (that I recall) eg:

https://nrsp.org.pk/hand-operated-and-solar-powered-portable...

https://www.bondheshams.org/

I read a piece about how abundant solar in rural Afghanistan strengthened the Taliban as poppy and therefore opium production skyrocketed because they were able to install electric wells.

interested in the article. would you like to share a link?

https://www.bbc.com/news/science-environment-53450688

Here in W.Australia we've been using solar powered signal repeaters along rail lines and mesa tops since the mid 1970's in the Pilbara mining industry.

Solar powered bores for farms in the wheatbelt and cattle country has been a regular feature for over a decade now.

The panels and batteries are better than ever now, but they've been Total Cost break even better than fossil fuel alternatives in remote Australia for a very long time (esp. factoring in need to travel to a bore to refuel, etc. Vs "it just works" (until it doesn't, but all solutions require upkeep)).

> Cost of renewable electricity in remote area becomes cheap enough to pull oxygen from air or electrosis from water to be sold as commodity

Oxygen is already kind of a commodity. Did you mean hydrogen? Anyway, this isn't going to happen, look at retail power prices.

> Top search engine results completely inundated by LLM text, we none the wiser. SEO solved. Search engine switches to new algorithm based on LLM.

I'm not sure in what sense SEO is "solved" here, but this basically puts the Internet into a nasty local minimum where it's kind of good enough for information you don't care about, but will feed you wrong information often enough to cause problems, while having stripped all the traditional reliability signals. Possibly also kills what's left of fact-based journalism.

I’m grateful for my librarian.

> 90% CL: Top search engine results completely inundated by LLM text

Are you sure you meant this as a prediction for 2025? It has been months, if not years, since search engines have been inundated by LLM articles.

The pursuit of advertising dollars has even made many long-tail queries worthless. Results often are a pile of AI-generated spam sprinkled with display ads and biased affiliate links. Ad blockers can deal with some of that, but there remains the problem of not finding appropriate results to many search queries.

This has been the sad state of the web for quite some time, and it’s just going to worsen.

I don’t think global temperatures get another record because El Niño is over. Your „weather“ suggests a more local reference though?

It sounds more clever to use Confidence Level, rather than just saying you are pulling random probabilities out of your a$$

Really bullish on AI, huh?

Designing an engine?

“[snuffs] out common diseases”?

Solving SEO?

Idk cotton, let’s see how this plays out.

Sniffs out, meaning diagnoses

Could just as well be a typo.

I’ve never heard anyone call diagnosing “sniffing”

It’s an idiom, like the detective sniffed out some clues.

lol, I do mean a sniff with a nose. The technology I'm referring to can work by taking air sample. This technology exist, the prediction is that cheap version will be available. My assumption is robots will want a digital nose, for example nanny bots will want to smell smoke and react. An example I read for this technology is that grocery store robot will want to detect spoiled food. It will be mass produced and gets cheaper. But we don't have a robot industry yet, so my confidence level on this prediction is low. The AI part of this is just data analysis of spectral lines.

Disease can change body odor, and dogs have been trained to "sniff out disease" [0][1] (i.e. detect via smell) for years. Various researchers have been working on robot noses that would be able to do the same.[2] Presumably a widespread and/or less expensive version of this "robot nose" is what doku means by "AI nose sniffs out many common diseases".

[0] https://www.nature.com/articles/d41586-022-01629-8

[1] https://www.discovermagazine.com/the-sciences/how-do-dogs-sn...

[2] https://bigthink.com/the-future/robotic-nose-smartphone-dete...

"Solving" is a bad choice of word. SEO does not have a solved state.

Digital design of engines can already happen now. AI chip design is already a thing, I think they help lay out more efficient trace lines. Since digital design and simulation can all happen digially. The whole cycle is digital. A system like Claude's computer use can take control and produce 100s or thousands of small modification of existing design to be simulated and if it has testable performance critiera. I think it can be done now. Maybe some finetuning the model for specific UI control.

I think it’s far more likely that a specialized model would be make for a such a task. Just like with chip design.

I’d be surprised if they weren’t already using some kind of ML/statistical analysis in the design of engines.

When I read AI, though, I was assuming you meant LLMs, not one of several statistical modeling technique.

How do those confidence levels work? It looks to me you're just assigning probabilities to outcomes.

I think it's roughly: "I would bet 9$ to your $1 that this happens", although some of these are vague enough that you couldn't objectively determine whether the bet materialized.

So they aren't confidence levels. They're odds, but he's trying to sound sophisticated.

can i have some of what you have been smoking?

I did my top 5 for my field cyber security

AI in cyber defence and attack Quantum resistant cryptography Death of the password? Passwordless, passkeys, phishing resistant MFA More suppliers compromised, another major open source project long con Doing more with less: automation and consolidation

https://rakkhi.substack.com/p/cyber-security-predictions-for...

Those are generic enough you could argue they already occurred last year or this year?

There is a lot more detail in the article that makes not generic

They have both occured and not occured every year for the past 10 years

Reply on news.ycombinator.com