inboulder

"shouldn't the Agent have succeeded way faster/efficiently instead of using "4.9 million messages and used about 300 billion output tokens."" No. Have you seen LLM output? The fact that it managed to converge at all in the problem space is amazing, an LLM is a 'book', the context window is the page number, the page has a token, that's it, that this huge book of weights already contains a method to search a problem space at all and actually go down a 'good' path eventually is remarkable, but it's going to be incredibly inefficient at doing so, a book doesn't have any kind of real-time model building like a brain.

LLMs are an excellent tool for discovering which problems mathematicians thought were interesting, actually aren't. If the LLM's haphazard jaunt (somewhat exhaustively) through the problem space can 'solve' it, the solution was never going to be particularly useful or illuminating.

PageRank for better centrality seems neat, but it still doesn't address the probably unsolvable flaw with RAG, the reason why RAG basically can't work. All RAG DBs under-perform expectations because RAG fundamentally can't find relationships between words necessary to find the information the user cares about. Weird right, isn't this what the 'attention' mechanism is supposed to be good for? It just isn't good enough.

Example: Say you're searching an article and you want to know what occupation a mentioned person has, let's say the person 'Sharon,' is mentioned to have attended several physical chemistry conferences but her occupation is never explicitly mentioned. There's a very good chance every single rag approach will fail to return correct results, will fail to make this connection between 'occupation' attends conference, type of conference and infers 'chemist'. I could go on, but this sort of error is pervasive along all types of information when trying to retrieve with RAG. In the end, solutions like the above seem to just sort of reinvent other query methods, SQL, pagerank etc, with extra steps... there's little point in vectorization at that point...

I think this article makes a fantastically gross error, the conclusion is fine, however the authors gloss over the fact that any space-faring life has a high probability of being machine-intelligence, not squishy biological life which 'evolved in the same manner as on earth'. (note, this is already the trend with our space exploration)

You're neglecting many reasons extra terrestrial life may pose a threat.

The most obvious is they may want our planet, they want to live here, and if they've managed to travel across the vast void of space, wiping out humanity probably won't be particularly challenging for them.

Also, your second point is invalid, territorial reasons are not void. We may pose no threat now, however a technologically superior civilization may conclude we pose a future threat, and so wipe us out now before we can become one.

Also, the old 'dangerous biology from somewhere else!' is almost entirely null and void, organisms are adapted to their environment, they usually pose negligible risk to any other environment already inhabited by biological organisms. This is why, for instance we don't treat organisms from thermal vents as 'oh noes, biohazard material'.

Can you point out the way in which the "Google Wave Bots & Gadgets" had meaningful elements of AI?

I don't think you understand my point at all. The pageviews Cheezburger garner ARE impressive. That it is possible to run a business with very low capital costs (based on wordpress blogs) fairly inexpensively (and thus grow or _shrink_ organically) is NOT impressive.

Sorry, don't get me wrong, I understand it's easy to burn through cash.

My point is: I'm not sure why Cheezburger thinks it's impressive to set up Wordpress blogs of other people's images without breaking the bank.

"Cheezburger has been a profitable business since inception. We have been profitable every single quarter, and never had a quarter with negative cash flows. "

Oh come on now, how could you have negative cash flow? Your whole business model is based around publishing content on the web you didn't create, (and 99% of the time probably don't even own the rights to), this is not exactly an expensive business to run.

>> For example, Java, can you tell me the difference between volatile and synchronized as it regards the memory model without looking it up? What's a happens-before relationship and how is it relevant to those keywords? What's the transient modifier do?

>I didn't claim to be a Java guru. Very few Java programmers actually need to know the answers to those questions in day-to-day programming; I could probably (if asked to, which I'm not, thankfully) write thousands or tens of thousands of lines of Java without ever needing to know the answer to those.

You're wrong, and this proves the point. This isn't 'guru' level, the workings of volatile and synchronized is even on the basic SCJP test. This may not come up in your domain, but if the project deals with threading, an experienced dev will be more productive than you right off the bat, after two weeks, and probably after a year.

In fact, threading is pretty tricky, so it's likely you will be counter-productive to the team until you've made all the Java threading mistakes you need to make in order to understand why your code is bad. And, you might never know your code is bad unless a domain experienced Java dev explains it to you.

Domain experience matters.

"Ernest Hemingway than Kurt Vonnegut"

I think you're trying to make a 'smart' reference but are confused. Overall, Vonnegut was generally at least as, or more concise; there is an entire page in _For Whom the Bell Tolls_ dwelling on the sweat on someone's brow.

Hey, I've been looking for a URL for my new app, this one looks perfect: beeftard.com Domain Name Sale or Trade ca98am79 4 hours ago

$300 is 15 lawns mowed, or 2k cans collected, or probably one post on your facebook page asking for donations.