Artisian

joined 3 years ago
[–] Artisian@lemmy.world 3 points 1 week ago* (last edited 1 week ago)

I read the article as making exactly your point. Agreed.

 

In any discussion about AI’s potential, it’s important to separate the technology from the socio-political system it’s embedded in. That AIs can lack context, mix up facts, or fall for stupid tricks are all technological problems. Because the giant developers like OpenAI and Anthropic have prioritized solving them, AIs can now more easily access resources like the web or email, are more disciplined about using those resources, and are better at staying within their guardrails.

Yet AI developers do not seem to be prioritizing other technological problems. Major AI models still act far more sycophantic than humans, telling people what they want to hear even when untrue or not in their best interests. Popular AI models tend to answer questions confidently even when they lack training, knowledge, or evidence to back their claims. In both cases, AI developers choose to train models that please users with flattery and the appearance of competence, rather than constraining them to act in users’ and society’s best interests.

...

In a different political and economic system, Chinese developers are producing—and then giving away—smaller, more efficient, more affordable models. While the US government seeks to restrict China’s access to the most advanced chips, China is betting that incentivizing their tech giants to create leaner, more open models using more commodity hardware—models that can be trained with older chips and run even on personal computers—will be an advantage in achieving widespread use and, perhaps, Chinese national influence.

There's a green powered, ethically trained LLM out there: Apertus. But we're not (currently) choosing that route.

[–] Artisian@lemmy.world 3 points 4 weeks ago

I do wish this was an easier sequence of steps to follow. I know I should, but it's so very much work.

[–] Artisian@lemmy.world 30 points 4 weeks ago (2 children)

Imagine. Oracle not being relevant. The dream

[–] Artisian@lemmy.world 4 points 1 month ago* (last edited 1 month ago) (1 children)

(page 53) The next line of that interview quote:

Eight of us managed to do it. The next time we attacked, we could jump.

And the footnote:

Interview with ISWAP Commander-7, 2025. While this may seem extreme, the episode reflects a pattern evident across other interviews, in which the group absorbed heavy casualties during dangerous experimentation and training.

It sure is wild; but humans aren't the expensive thing for Boko Haram I guess?

[–] Artisian@lemmy.world 4 points 1 month ago

specialists worked side-by-side with the tools, consulting the chatbot in direct support of technical work, as in the recounted case of a bombmaker who iteratively refined new device designs with AI, returning to the laptop for troubleshooting when something was not working or needed clarification.

Regarding the first usage, both factions seize a significant part of their weaponry, military equipment, and ammunition from the Nigerian military. Struggling to operate some of the loot, ISWAP’s leader al-Barnawi stated in a letter to the Islamic State back in 2017: “I am reminded of a funny story when about a year ago, we seized Dragunov rifles as war spoils and we had to make them work so we could use them. We checked many handbooks and specialized videos on this technical issue, but we eventually dropped it in despair after all our attempts failed.”88 This account contrasts with how these situations are handled and resolved today, where “we take all of the equipment back to the camp and the AI unit tells us how to use and shoot with it.”

Perhaps not surprising that it performs well on these sorts of tasks, despite what they say the military is a big buyer....

[–] Artisian@lemmy.world 7 points 1 month ago

Back in my day, we memorized logarithm tables and we liked it!

[–] Artisian@lemmy.world 1 points 1 month ago (1 children)

Do you have a tutorial on how that's done?