techwiz137
techwiz137 on Hacker News
For code generation, nothing so far beats Opus. More likely than not it generated working code and fixed bugs that Gemini 2.5 pro couldn't solve or even Gemini Code Assist. Gemini Code Assist is better than 2.5 pro, but has way more limits per prompt and often truncates output.
I do not particularly agree with this. I don't think this analogy works.
I had a heart attack moment thinking they were bringing some form of Adobe Flash back.
Photons, held by force-fields? Sounds Trekky.
I like the taste. This is why Monster energy and Coca Cola are my only two drinks. With a hint of 7-up.
I attempted this just recently, I went 12+hrs without caffeine intake and I had the worst, headache that wouldn't go away with regular pills. Basically very intense caffeine withdrawal.
As far as I've read, our coke zero has a lot of added phosporus or a source of, it seems to somehow bind to calcium leading to perhaps bone loss.
I had an injury I thought was unrelated, but could be. Premature ankle joint wear. On inspection from a CAT scan machine, I was told I seem to have some abrasions on the bone around the damaged area, likely from or similar to osteoporosis and/or osteoarthritis.
I can be a case study. I drink over 2 liters of Coca-Cola with aspartame, per day, every day on an empty stomach. With an energy drink, containing aspartame. I am not bragging, it is an addiction.
Might be true, but I have a form of dyscalculia that made almost any sort of math impossible for me, so for this it doesn't affect me as I am not capable either way.
But AI has taken away my ability to think to solve a problem via code. Now all I do is just ask it.
I am fairly certain the hard drive is beyond repair. It has more than likely had water ingress by now.
That is exactly what has happened.
How do they know the internal microcode structure of instructions or even format?
How can you have a bit and a half exactly? It doesn't make sense.
I think given the advent of AI, there is less of a need of such a language. All I need is to type out a piece of concise text to AI of choice and I can have a general GUI up and running without having specifically much knowledge of a programming language.
You are quite right, and not only can anyone write anything, but you have a double whammy from the LLM which can further hallucinate from said information.
The biggest question is, can this bring back the behaviour of search engines from long ago? It's significantly difficult to find old posts, blogs or forums with relevant information compared to 10-15 years ago.
My car runs on methane, but it's very expensive, only 20% cheaper than gas and soon it might be 1:1. Hard to store (200 bar pressure tank) and tanks have a 20 year lifespan.
Interesting, I have always noticed a pattern in my social experiments whereby a topic about AI complimented with AI generated content always results in a negative sentiment. Like a feedback loop.
Since it is a big wall of text, let us ask the subject to summarize. Let's see if it did a good job.
"The article "GPTs and Hallucination" by Jim Waldo and Soline Boussard explores why large language models (LLMs), like ChatGPT, generate hallucinations—responses that are realistic but factually incorrect. These models, based on massive data sets and probabilistic predictions, are powerful at generating coherent text but aren't anchored in real-world facts. Instead, they predict word sequences based on past language patterns, not objective truth, which leads to hallucinations when there isn't enough reliable data or consensus on a topic.
The article describes how LLMs are trained, noting that their responses are derived from patterns in the data, not semantics or factual accuracy. When there’s widespread consensus on a topic, LLMs tend to give correct responses, but when the subject is obscure or controversial, hallucinations are more common. The authors conducted an experiment using different LLMs (Llama, ChatGPT-3.5, ChatGPT-4, and Google Gemini) to test this hypothesis. Their results showed inconsistencies, with more advanced models like ChatGPT-4 providing more accurate responses, although hallucinations still occurred, especially with obscure or controversial prompts.
The authors argue that LLMs mirror crowdsourcing, relying on consensus from large language data sets. However, while this works for common topics, LLMs struggle with niche or polarizing subjects, highlighting the need for caution when using these tools for critical decisions. The conclusion emphasizes that LLMs perform well on popular topics but are less reliable when the data is sparse or contentious."
In what update is this packaged for Windows 10 and Ryzen 7800x3d?
In my country we have a saying. "Porcupine in the pants". Sounds like for all the good it can do, it isn't written safely and carefully.
The poster is saying that cryptocurrency is a pyramid scheme. 12 years ago(or 13) in its infancy, you could've had this sentiment.
But now you have ETFs, it's as legitimate as it comes.
This again? I can understand the sentiment 12 years ago, but now?
How about people moving away from 16-bit real mode(obviously a change is needed in the CPUs), removing that A20 line forgotten from history patch and actually booting like we are supposed to?
In ICE cars, some people remove the mufflers and modify the exhaust for louder and more deeper sounds.
For the time it took them to develop it, the performance is really lackluster. I would've definitely wanted an OrangePi 5 equivalent performance, but with the mature software of the RPI.