fsndz

12 ms·

151.
by fsndz 2y ago
Building robust LLM-based applications is token-intensive. Luckily, the cost of intelligence is quickly dropping.
152.
by fsndz 2y ago
If it's the H1B, why attack another person so viciously instead of attacking the system? By the way, it's the same story for immigrants almost everywhere. (African immigrant in Europe here—it's even tougher: companies literal
153.
by fsndz 2y ago
This is basically the LLM modulo approach recommended by Prof. Subbarao Kambhampati. Interesting but only works mostly for problems that have some math/first degree logic puzzle at their heart. Will fail at improving perf at ARC-AGI fo
154.
by fsndz 2y ago
wow, who hurt you ?
155.
by fsndz 2y ago
True. In the classic form of automation, reasoning is externalized into rules. In the case of AI agents, reasoning is internalized within a language model. This is a fundamental difference. The problem is that language models are not design
156.
by fsndz 2y ago
I think they look at the overall impact not just one single thing. a lot of people invented backpropagation independently but after that their impact was low. Hinton is like everywhere when you look at the state of machine learning now
157.
by fsndz 2y ago
absolutely. the pervasiness will only continue
158.
The impact of generative AI as a general-purpose technology(mitsloan.mit.edu)
2 points by fsndz 2y ago|2 comments
159.
by fsndz 2y ago
The battle of tactical intelligence
160.
by fsndz 2y ago
fair enough
161.
AI Can Only Do 5% of Jobs, Says MIT Economist Who Fears Crash(bloomberg.com)
9 points by fsndz 2y ago|2 comments
162.
by fsndz 2y ago
this is true in theory but not in practice. just read the article, they came up to the same conclusion. if you want to offer a slick UX (a key element to convince users to pay these days) htmx won't cut it.
163.
by fsndz 2y ago
convenient. I would have liked taking a look at your apps built only using htmx
164.
by fsndz 2y ago
sorry for the paywall, you can read the free version here: https://www.lycee.ai/blog/why-no-agi-openai
165.
by fsndz 2y ago
I will start using htmx as soon as there is a 20K MRR startup built only with it.
166.
by fsndz 2y ago
openai is basically eating all the GPT wrappers over time, as well as integrating all the interesting work done at prompt level (cot, structured outputs) at the model level. I wonder if that trend is doomed to continue as profitability issu
167.
by fsndz 2y ago
the sexual assault allegations seem bogus to me
168.
by fsndz 2y ago
LLMs are basically becoming commodities: https://www.lycee.ai/blog/why-large-language-models-are-a-co...
169.
by fsndz 2y ago
Exactly what I wrote recently: "The "AI effect" is behind some of the current confusion. As John McCarthy, AI pioneer who coined the term "artificial intelligence," once said: "As soon as it works, no one calls
170.
by fsndz 2y ago
Debates about artificial intelligence are often semantic debates. Artificial intelligence is not "far from existing," but artificial general intelligence (AGI) is far from being achieved (large language models still do not imitate
171.
by fsndz 2y ago
> "In the early years, getting AI models like ChatGPT or its rival Cohere to spit out human-like responses required vast teams of low-cost workers helping models distinguish basic facts such as if an image was of a car or a carrot.&
172.
by fsndz 2y ago
this is not how GDP works
173.
by fsndz 2y ago
crazy that this is true and GDP just keeps increasing anyway
174.
by fsndz 2y ago
"Doing work" vs. "performing work": the epitome of this is consulting. Companies pay huge sums of money to consultants that often spend most of their time "performing work", doing beautiful slides even if the c
175.
by fsndz 2y ago
this is a top notch comment !
176.
by fsndz 2y ago
there is so much junk science these days and the problem is the incentives are wrongly set (quantity over quality)
177.
by fsndz 2y ago
I'm willing to bet that if you tried Rails, it probably wasn’t a good fit for you.
178.
by fsndz 2y ago
So true, and this is a dilemma, right? People who build frameworks do so to make it easier for others to ship products. In the process of building the framework, they become better developers themselves. However, others now have to learn th
179.
by fsndz 2y ago
how do you manage when the core concept are too abstract ? I guess then you would need some examples to understand ?
180.
by fsndz 2y ago
A new language model programming library has been released, and it’s called Ell. It aims to compete with, and possibly replace, libraries like LangChain, Llama-Index, and DSPy. I know — yet another library for composing language programs. A

More ›