jsemrau

10 ms·

121.
by jsemrau 2y ago
The same timing is actually suspicious. And it would not be the first time something like this happened.
122.
by jsemrau 2y ago
DALL-E 3 already uses a model that trained on synthetic data that take the prompt and augments it. This might lead to the overfitting. It could also be, and might be the simpler explanation, that its just looks up the right file from a RAG
123.
by jsemrau 2y ago
>Apparently there's some 7 million young men of working age that are...missing in action. Self-isolated, gaming, addictions. And you never wondered why that is?
124.
by jsemrau 2y ago
I suppose that's the problem I have with that study. T
125.
by jsemrau 2y ago
Great point
126.
Nemotron vs. Qwen – Game Theory and Agent Reasoning IV(jdsemrau.substack.com)
1 points by jsemrau 2y ago|0 comments
127.
by jsemrau 2y ago
I think the key takeaway quotes are these: “The amount of inference compute needed is already 100x more” than it was when large language models started out, Huang said on last month’s earnings call. “And that’s just the beginning.” The cost
128.
by jsemrau 2y ago
Products are developed for a specific market. If the market of the given minority group where large enough, there would be special providers only for that. If I go buy an Italian Pizza place, I expect to receive an Italian Pizza, and not an
129.
by jsemrau 2y ago
“I have studied, alas, philosophy, jurisprudence and medicine, and even, sad to say, theology, all with great exertion.” – Faust, Part I, Scene I (“Night”)
130.
by jsemrau 2y ago
(1) Daemons run usually in the background while agents are main loop. (2) Daemons are predefined and can't adapt to changes easily -- this is a function of narrow AI vs general AI (not to be confused with AGI) (3) Daemons have few inte
131.
Chinese scientists build first AI chip made of carbon and its super fast(scmp.com)
2 points by jsemrau 2y ago|0 comments
132.
by jsemrau 2y ago
Thank you!
133.
by jsemrau 2y ago
"where agents would ever be better than workflows" That is a very important observation and we should avoid to let agents go the way of the blockchain -- you know what I mean. I have build a narrow AI for credit decisioning on a 1
134.
by jsemrau 2y ago
Add this please -> https://jdsemrau.substack.com/about
135.
by jsemrau 2y ago
In some cases that is intended behavior. Also look at block trades - https://www.investopedia.com/terms/b/blocktrade.asp
136.
by jsemrau 2y ago
If only Google would stop indexing social media and start linking to blogs and sources of real people the world would be better off. Maybe we need a humanity score for the search results and not an engagement score. Do I really need to see
137.
by jsemrau 2y ago
No they are not. Here is an example how you can buy a NIO in Germany https://www.nio.com/de_DE/get-your-nio/kauf EV is on a decline in general after the end of government subsidies And Germany is still dominated b
138.
by jsemrau 2y ago
I would expect that this exactly what the people who voted for the current government wanted. I would like to learn more what they were researching, though.
139.
by jsemrau 2y ago
"mouthpieces for the ruling class" That's actually a great point. Judging from the current state of media, there is a clear momentum to take sides in moral arguments. Maybe the standard for models need to be a fair use clause
140.
by jsemrau 2y ago
Is it the same thing though? Even though Lord Of The Rings, the book, likely has been used to train the models you can't reproduce it. Nor can you make a derivative of it. Is it really the same comparison like "Simba the white li
141.
by jsemrau 2y ago
Of course I read books by women. I would only encourage you to research if the blood, sweat, and tears of your ancestors are the "socio-political effects of European patriarchy" or just people who build a home that attracts people
142.
by jsemrau 2y ago
For cloud providers it makes sense to be model agnostic. While we still live in a datacenter driven world, models will become more efficient and move down the value chain to consumer devices. For Enterprise, these companies will need to reg
143.
by jsemrau 2y ago
Yes, but market making was difficult.
144.
by jsemrau 2y ago
Correct. OpenAI Search, Perplexity, et al are Answer Engines and not Search Engines.
145.
by jsemrau 2y ago
https://news.ycombinator.com/item?id=40133976
146.
by jsemrau 2y ago
What would a product look like in this space?
147.
by jsemrau 2y ago
I pay (and write paid) on Substack for content. There are good people who provide unique insights. I think I prefer to live in this world than in an ad engagement farm.
148.
by jsemrau 2y ago
Content marketing died already many years ago.
149.
by jsemrau 2y ago
Wasn't poe.com supposed to be the quora replacement since its run by the same guy who also sat on the OpenAI board?
150.
by jsemrau 2y ago
Interesting observation.It seems capability has reached a plateau. Like a local maximum.

More ›