AI is your new master ....

[byates1]

https://youtube.com/shorts/9sdvaLpye...O19z5tCsrH8EYR

[byates1]

https://youtube.com/shorts/eV7DYlbM8...U21rKN-Mj6h4tG

[byates1]

This guy's vids are great

Sent from my Pixel 8 Pro using Tapatalk

[Kevo]

Deep seek R1 looks like it could be an incredible advancement at a fraction of the price and it is open source.

Huge implications for closed LLM companies.

[Mudfoot]

https://www.youtube.com/watch?v=VZuMcqI9xoo

A download of the book this is based on is available for free in PDF form.

[Meadow Skipper]

^ Well, that was dark.

[ötzi]

Oh I'm sure it'll be fine.

[JimmyCarter]

“AI will most likely lead to the end of the world, but in the meantime there will be great companies created with serious machine learning”
- Sam Altman, maybe tongue in cheek

[J. Barron DeJong]

Holy shit. Economist asks DeepSeek “ Create a model and run a simulation to show the impact on Canadian GDP if the US imposes 25% tariffs on all exports from Canada to the U.S.” and in 12 seconds it has a pretty thoroughly reasoned argument, model, and result inline the estimates from the Bank of Canada. (I’m underselling it, click through to see how thorough it is from a single prompt.) https://stephaniekelton.substack.com...utm_medium=web

[J. Barron DeJong]

Well, there are limits. I started asking DeepSeek about race car suspension design, which I know a fair amount about, and it quickly started hallucinating (you’re correct to question this! I misspoke!) so work to be done still.

[jono]

Not that I'm not happy to see that, but could we not train the AI on these particular specialities just yet? TIA

[old goat]

DeepSeek is thousands of Chinese people sitting at desks looking up the answers in Wikipedia. You know, the Theranos model.

Here's a question--I'm obviously completely ignorant on the subject--could you develop an AI app by training it on someone else's AI app? How many times does all the knowledge in the world have to be scraped bit by bit?

[Not DJSapp]

^^^ Deepseek is pretty much that. It's a lightweight AI that trained from the heavyweight AI's. Because it's standing on the shoulders of the big ones, it can do a lot on low spec hardware, but it is prone to issues if you push it. Here's a dude who seems to know a lot more about it.

[shroom]

and others will continue to add abstraction

[mcski]

at first I thought this was an Onion headline. Lmao, someone please sue ChatGPT for copyright infringement and every other ip law possible

[MultiVerse]

Got full main DeepSeek-R1 up and running locally in less than a day. This Chinese soft-power psyop is pretty good, like having several hundred billion worth of stolen ChatGPT IP sitting in a box in my basement doing crazy godlike things

[shroom]

I bet you clone your pets and ride a hoverboard to work

[MultiVerse]

heh, we were promised jetpacks and all I got was this otherworldly intelligence

[Rasputin]

Speaking of otherworldy intelligence, I hadn't heard of this before.

https://www.youtube.com/watch?v=ivB2QUxy3S8

[jono]

You and Michio Kaku both.

[Rideski]

I just did an AI job interview. It was way more advanced than I would have presumed. Asking rapid fire complex two part questions. And then taking my comprehensive answer and reframing it and speaking back to me in less than a second. This went on for half an hour.

[PB]

Who was the interviewee, and who was the interviewer?

[Clownshoe]

I fiddled around with the app for a couple of hours. I asked it some questions about stuff I know a fair bit about, and it quickly started very confidently giving me incorrect answers. Basically exactly with GPT does.
So my takeaway from this is the same as with GPT. Cool, but generally "meh" for my applications.
I'm clearly not using it to its full potential, but when it seems like the these apps can't even summarize a google search properly, I'm pretty suspect of answers it gives me that I can't check myself.

I do like the Instagram AI's ability to create images. I've definately killed some time getting it to make photos of me skiing with my cat while volcanoes erupt and explosions go off around me

[MultiVerse]

For what it's worth, an LLM can be trained with propriety data for targeted usage

[byates1]

https://youtube.com/shorts/mAZU0RLy0...3IK5Psh9pgfLgX

Almost looks real. Couple yrs forget it.