MVissers

If needed and wanted methane can be created sustainably. But not sure about upper atmosphere issue.

Most average humans can’t do any of these things either. Try asking people on the street. Or in an average US college student.

For deterministic calculations you obviously want to allow LLMs to use tools to do math. Just like you’d want to allow humans to use calculators.

So yeah, you shouldn’t ask LLMs to do math just like you shouldn’t ask average people to do math. They both suck at it.

They are starting to be smarter at both analyzing images and speech as well. They’re still behind on simple reasoning (eg. O1-preview), but it’s catching up quickly.

Obviously these models still have trouble interfacing with the real world.

Which model? The field moves so fast it’s hard to validate statements like this without that info.

O1-preview?

An evolutions simulator where both plants and animals can evolve both their behavior and morphology.

This month I’m focussed on plants and I think I’ve had a breakthrough that could be worth publishing.

But just a hobby.

I’d say depends. If the model iterates 100x I’d just say give me the output.

Same with problem solving in my brain: Sure, sometimes it helps to think out loud. But taking a break and let my unconcious do the work is helpful as well. For complex problems that’s actually nice.

I think eventually we don’t care as long as it works or we can easily debug it.

Doc here. It’s hard to prove.

You can’t do a double-blind randomized control trial.

And we have barely learned to measure these nanoplastics.

But there are starting to be strong correlations between the plastic occurrence and terrible diseases. Recent data on atherosclerosis + infarcts is horrible.

Recent data on testes production of sperm cells also looks very damning.

In summary, there are a few unexplained things in medicine that might very well be caused by plastics. Eg. More colon cancer in young people in the west. Eg. Declining sperm quality in the last 30 years in the west. Yes, it could be foods, but it doesn’t explain the whole picture. Plastics could be it.

My guess is that a LOT of nasty studies will come out in the next years.

I switched to glass and metal for all my food and water, for as much as I can control it.

3 top people leaving in one day. This is fine.

No, look at this: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6455044/

Decreased sperm count in the west. Could be due to a variety of factors.

But tie it together to microplastics: https://pubmed.ncbi.nlm.nih.gov/38745431/

And you got an interesting result.

Humans and pet dogs: https://academic.oup.com/toxsci/article-abstract/200/2/235/7...

Fear mongering?

Most renowned medical journal would disagree my friend: https://www.nejm.org/doi/full/10.1056/NEJMoa2309822

4-5 times more chance to have another heart infarction in the next year.

Paper on decreased sperm count over time in the west (prob many reasons, plastic could be one) https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6455044/

These things are very bad. We're just starting to learn HOW bad they are.

Thing is- We’re only now starting to realize how bad plastics really are, because we only recently acquired the tech to measure this.

Unexplained fertility decline in the last 40 years? Probably plastic. They found that humans had 3x higher concentration of plastics in their testicles than their pet dogs, and that pet-dogs have a dose-dependent reduction in viable sperm count inversely correlated with the plastic concentration.

Have a plaque in your blood vessel body with high plastic concentration? Since this year we know it has a 4x (!!!) increase in a new heart infarct in the next year compared to when there is less plastic.

It’s starting to show that it is really bad, and it’s virtually everywhere in our body, most likely causing a ton of issues.

You can use LLMs to vet the relevancy of the content, so you only select the most useful data. I believe most labs are doing this today.

Absolutely for learning. If you want to learn something we should realize it's awful. Try to learn something in your own field to see how awful it is.

That's why we're all armchair experts in other domains.

Also better access to healthcare.

I’d guess it helps short-term survival.

Eg: Rest your twisted ankle through pain and inflammation instead of keeping to use it.

Even though it increases recovery time, better be on the safe side of decreased usage to avoid further damage.

Which is why in modern times RICE speeds up recovery (rest, ice, compression, elevation) in twisted ankles.

But we have ratio- which mammels didn’t have for millions of years.

Agree with this. It’s the most beautiful night sky I’ve ever seen.

The stars seemed to shine 10x brighter than other places, even those without light pollution.

Sounds like a tremendous step backwards, since congress is directly paid for by mega corporations.

Especially concerning complex topics such as EPA and FDA. People (voters) don’t understand these topics and neither do politicians. So how can we expect politicians to regulate these complex areas?

Sounds like corporations will just start to own the laws that are being passed even more.

Which promotes further consolidation of power and detrimental health effects for the population.

So the FDA will lose power to regulate what type of food additives are allowed or not?

It makes sense IMO that orgs like the FDA or EPA regulates health safety issues instead of politicians.

Does this block that from happening?

Not a big issue with evs, most use regenerative breaking.

Could be related to high interest rates and investments being more expensive. Also, I believe that Airbus has a 10 year backlog on their newest planes, so no point in shelling out a lot of cash right now for 10 years down the line.

Apple is starting to anticipate a higher RAM need in their M4+ silicon chips: There are rumors they are including more ram than specified in their entry level computers.

https://forums.macrumors.com/threads/do-m4-ipad-pros-with-8g...

One reason could be future AI models.

I'm not sure if this has been verified independently, but interesting nonetheless and would make sense in an AI era.

They showed reduced sperm count in a dose-dependent manner in dogs. More plastic, less sperm.

And then a higher plastic concentration on average in human testis.

Wild.

LLMs don't think, they just regurgitate text tokens based on a probability + randomness model.

/s

Disney would sue them for that, no?

If it sounds too alike intentionally, it’s basically an impersonation of a character, which falls under copyright.

Someone else was sharing case law about this.

He’s so power hungry it’s kinda crazy. He’s also extremely effective at it, like a real machiavelli.

PG said the same- Never has he seen someone which such a capability to gain power.

Not OP, and not a computer scientist. But amateur neuroscientist.

From what I understand, current models seem to have issue with planning and reasoning because they “blurt” everything out zero shot. They can’t recurrently process information.

Part of it is how we train these models, which is rapidly being experimented with and being improved.

Another part is that our brain can both consciously and unconsciously process information recurrently until we solve a problem. Especially helpful when we haven’t solved a problem before.

With LLMs, we can do that rudimentary with recurrent prompting. In a way, it can see the steps it has tried, re-evaluate and come up with a new try.

But it’s not innate yet, where models can “think” about a complex problem until it solved it.

Also- According to some theories around consciousness, this is when consciousness will really emerge (integrated information theory).

I’m pretty sure we’ll be able to solve this recurrent processing issue in the next few years.

Not really. You can only make these predictions about things bottlenecked by things that improve exponentially such as compute.

Neither the space program nor self driving are compute restrained.

The latter will probably be “solved” as we get closer to AGI, since you need some sort of human like reasoning for edge cases that require reasoning.

Another tech like this is batteries: There is no miracle jump in production batteries. They just improve about 10% YoY, both in energy density and cost.

So you can extrapolate when electric cars will be cheaper than gas cars to buy.

Intersects around 2035 last I checked.

I mean, this new openai model is trained on video and voice as well.

So already, his argument is outdated. He had made a few of these claims over the last year that have been quickly disproven by a new model.

If you look at compute scaling and model improvement on that compute, we’re going to get there pretty fast.

Both compute, architecture and cost matter, and those have been improving like crazy.

I don’t know about you, but I didn’t expect GPT-4o to come out at half the price one year later, with real-time voice, image and text.

There is zero sign of slowing, Nvidia keeps building beefier GPUs specialized in LLMs, the world keeps adding more data centers and humanity keeps creating exponentially more data world wide.

You can plot this exponential growth out over time and calculate when these models will have the complexity of the brain. Then you can assume some penalty for shitty architecture (that gets better over time), and you’ll have a ballpark estimate.

Somewhere 2027-2030 we’ll have models that most of us consider AGI today. Now, societal impact is hard to predict since that’ll depend on regulators as well.

Tbh, he wasn’t convicted as far as I know.

But yes, he’s overly optimistic with timelines. He says so himself.