dissahc

because it wasn't trained to play chess

imagine a hypothetical chess match between:

- an undoubtedly very intelligent person. in the course of their studies, they have read about different chess strategies, openings, etc. but they never actually played the game themselves

- an average person with a year of chess playing experience

who do you think is going to win? of course, you could give the LLM time to think and consider its opponents potential next moves, but this is a computationally expensive way to play the game that doesn't scale

which is all beside the point that chess isn't a very good proxy for general intelligence. there is a correlation, but it's very weak

It's an incredibly smug poem that has the vibe of someone fantasizing about winning an argument against someone who doesn't even exist.

The narratives about people not calling their friends for advice, but instead using AI... these are basically unfalsifiable. How could he possibly know that this is a thing that commonly occurs? That's right, he doesn't.

>Be sure to use AI when making your next, I don’t know, meal plan, for example. Definitely do not call your friend who loves to cook and ask her for her favorite recipes or tips or ways to save time making meals

I don't know about you, but if a friend were to call me for meal prep advice, I'd honestly be worried that they're having some sort of crisis and just need to talk to someone.

>Definitely do not text your friend who has fly-fished every river in Pennsylvania and biked every backwoods trail

Personally, I consider it kind of rude to pester someone who is an expert in a subject with extremely basic questions. Yes, sometimes they wont mind and will even be glad to answer your questions, but they would probably appreciate that you took the time to do research and aren't just using them as human Google search. The more genuine way to reach out to this person would be to learn as much about the subject as you can, and ask them to sanity check what you've learned. This is a much more considerate way to go about things.

>Be sure to use AI

>and while you do I’ll be over here in my 50th

>year, my youngest daughter asleep on my chest,

>my arm falling asleep because I dare not move

>lest I scare away this moment,

What? Does using AI disqualify you from having this experience? This post is so ridiculous man.

qwen3.5 9b outperforms granite 4.1 30b by a huge amount (32 vs 15 on artificialanalysis benchmark)... i have no idea what made the writer of this article say so many demonstrably incorrect things

isn't object recognition essentially solved? AI models were beating humans at image classification (in terms of error rate) back in 2016. even if this particular model isn't the best at it, they can always call out to an API or have a secondary on-device VLM that has stronger object recognition capabilities

$899? The bladeRF 2.0 micro xA9 has identical or even better specs -- for $860. Aren't crowdfunded projects supposed to be a bit cheaper to make the gamble on a new product of unknown quality and performance worth it?

The bladeRF 61.44MHz sampling rate can be effectively doubled (although they both use the AD9361, so it principle also possible on the SignalSDR) and the FPGA on board has 301KLE, versus the SignalSDR 85KLE.

The comparison matrix is also being a bit dishonest by not mentioning that the bladeRF also supports USB 3.0 and can fully saturate the 5Gbps link -- no such information about the SignalSDR.

It also states the PlutoSDR frequency range as 325MHz-3.8GHz and RF bandwidth as being 20MHz, but it can easily be increased to 70MHz-6GHz and 56MHz RF bandwidth by running two commands on the device. This process is approved by the manufacturer, and is actually outlined on AD's official wiki for the PlutoSDR.

The crying shame is that none of this potential can be utilised due to the limitations of the USB 2.0 port. The oscillator precision is also lacking, but modifications are possible.

But what I'm trying to say is, the PlutoSDR is $200, sometimes on sale for $99. If you added USB 3.0 and improved the oscillator precision, it would be practically on par with the SignalSDR, and that wouldn't cost $700 extra to do.

What a terrible article. There was one part that may have actually communicated some information though

>But by using entangled photons, the quantum photonic-dimer laser can maintain precision and strength over greater distances and in adverse conditions, the scientists said in a statement.

This makes me think the "photonic-dimer laser" has very narrow linewidth/extremely long coherence length, but also at very high power. There's usually a big tradeoff between laser power and spectral purity. A laser with a longer coherence length has favourable beam characteristics which allow it to be less divergent and less susceptible to diffraction from the atmosphere. The ideal propagation characteristics combined with the high output power is the story I guess.

Quantum entanglement contributes to the longer coherence length, which I guess could described poetically as photons protecting each other...

are you sure you need a powerful model for this? llama3-8b is at least 10 times cheaper and might suffice for something like this

i don't think having weaknesses like these say too much about the general capabilities or potential of LLMs. try this yourself: generate a novel pangram without any sort of iterative process or revisions -- the first thing that pops into your head you must commit to paper. it's very hard. it's a lot easier if you go in alphabetical order, as you don't need to keep track of which letters you've already used. interestingly, gpt-4o also performs better at this task when you ask it to go in alphabetical order.

LLMs have known weaknesses, many having to do with an inability to "think" without using tokens. so for tasks like these, you can dramatically increase their performance by getting them to think out loud. this was the crux of the "let's verify step by step" paper.

i ran this prompt three times:

### prompt start ###

your goal is to create a novel pangram in the spanish language that sounds natural. it should be grammatically correct and coherent. it shouldn't be just technically coherent and grammatically correct, it should sound like a normal sentence.

first, print out the spanish alphabet. these are your $remainingletters. $sentence = "". then enter a loop where you do the following:

loop 1: - select a random letter from $remainingletters choose based on which one allows you to add the most natural sounding word to the sentence, do NOT go in alphabetical order - eliminate the letter you chose from the remaining letters - ensure the word you chose actually begins with that letter! very important - add the word you chose for that letter to $sentence - are there letters in $remaining? if so, go back to start of loop 1. otherwise move on to loop 2.

loop 2: - go through the spanish alphabet in order and ensure your $sentence contains a word starting with that letter - once every letter is accounted for, translate the sentence to english - does it sound like a natural sentence? - if not, go back to the start of loop 2 - if so, print $sentence as well as its translation in english

think out loud, keep track of your work as you go

i'm not asking you to generate code, i'm just explaining how you should accomplish the task

### prompt end ###

i ran this prompt three times, and each time it generated a valid pangram that was coherent and grammatically correct. i can't be bothered to run it a bunch of times to get an accurate success rate, but i'm fairly sure there exists a prompt with a success rate of 100%. there are a lot of output tokens available, and by asking the model to iterate, it will arrive at something correct far before 128,000 tokens are exhausted.

sidenote: when i asked gpt-4o to generate 5 novel pangrams in english, and it got them all right. so language definitely matters when it comes to getting things like this right in one shot.

>One exercise that helps every time is I’ll draw a diagram of all the major characters and ask myself, “Does each character have at least one scene with every other character?” Or “Am I missing moments with three or four characters together?”

this sounds like a great way to insert a pointless and boring scene into a TV show. why not just focus on making good television instead of servicing characters because... ???

they could contain synthetic nicotine

Yes it can, but how often do you see ChatGPT make grammar or spelling errors unless prompted to do so?

I think the fact that there are grammatical errors proves that it's not a confabulation.