huginn

joined 2 years ago
[–] huginn@feddit.it 16 points 8 months ago (3 children)

I heard rumors someone burned his house down. No idea if they're true or not.

[–] huginn@feddit.it 24 points 8 months ago* (last edited 8 months ago) (1 children)

PLA is widely used as a medical plastic and its normal decomposition is into lactic acid.

Even if it is just being atomized down into smaller and smaller particles it's safer for you than any other common plastic.

The colorants added are the only risk

[–] huginn@feddit.it 13 points 8 months ago

Yeah that's about the face I was expecting him to be making. It's the "Please knock my lights out" face. The mental agony face.

Working long hours is sometimes justified: but unless you're the CEO you should be getting extra pay or time off in lieu. Anything else is wage theft.

[–] huginn@feddit.it 1 points 8 months ago (2 children)

He has said that they already know everything they need to know to get to AGI.

OpenAI has not made a single thing that wasn't just a wrapped LLM.

So either A: he has somehow been running a skunk works that has fundamentally changed everything we know limits LLMs and none of the researchers leaked anything or B: he thinks LLMs are the way.

Additional quote when he was asked about AGI:

“How did we get to the doorstep of the next leap in prosperity? In three words: deep learning worked. In 15 words: deep learning worked, got predictably better with scale, and we dedicated increasing resources to it,” Altman said.

This tracks with the whole "I need 1 trillion dollars of energy investments" plan to get to AGI that he's asked for.

The guy's either honestly convinced himself that we can get there with deep learning scaling or he's a conman. Could be both.

[–] huginn@feddit.it 0 points 8 months ago (8 children)

It's what Altman has constantly said was going to happen. Up to you to decide if he's actually in the industry or not.

[–] huginn@feddit.it -3 points 8 months ago (1 children)

The fundamental architecture of an LLM is used across all these services - and that's reaching its limits.

[–] huginn@feddit.it 4 points 8 months ago (1 children)

Heavy investments is a strong term for modest hedges around SMRs.

Tens of millions is low risk pocket change compared to the billions burned running the things constantly.

The problem is they're all playing chicken with each other.

OpenAI will never back down. The question is will Google, Microsoft or Amazon blink first

[–] huginn@feddit.it 4 points 8 months ago

My point is that it was foreseeable and it was her goal. She never wanted congestion pricing, she wanted to run on it because it was popular and then tank it in a way that won't hurt her reelection.

This is calculated.

[–] huginn@feddit.it 9 points 8 months ago (2 children)

She's moving forward with it so that Trump can block it. Then she'll say "look at me fight trump" and also "nothing we can do sorry"

[–] huginn@feddit.it 4 points 8 months ago

I'm one of the new AMD users as of yesterday. Took some fiddling around but I've got my new machine up and running. I had issues with my i9 being unstable and no patch was going to fix it. Pretty sure it was physically damaged from the manufacturing defects.

So AMD it was. So far so good!

[–] huginn@feddit.it 17 points 9 months ago (6 children)

I'm pro nuclear as well but we absolutely can maintain this level of energy consumption on renewables alone.

The question is cost and risk - I'm for diversification of our grid which includes nuclear.

But it is getting to the point where renewables with backups will be cheaper than coal. That's absolutely something you can run the entire grid off of. You can balance storage requirements with excess production capacity that gets shuttered over the summer etc etc

[–] huginn@feddit.it 10 points 9 months ago

Quote from abstract:

We find that the MTEs are biased, significantly favoring White-associated names in 85.1% of cases and female-associated names in only 11.1% of cases, with a minority of cases showing no statistically significant differences. Further analyses show that Black males are disadvantaged in up to 100% of cases, replicating real-world patterns of bias in employment settings, and validate three hypotheses of intersectionality. We also find an impact of document length as well as the corpus frequency of names in the selection of resumes.

Pretty damning... And not surprising.

view more: ‹ prev next ›