It's easy to believe if it's 5x $200 subscriptions.
Paying by the token is insanely expensive. Only the
5̵ ̵R̵i̵c̵h̵e̵s̵t̵ ̵K̵i̵n̵g̵s̵ ̵o̵f̵ ̵E̵u̵r̵o̵p̵e̵ Biggest Tech Co's can afford that.
But the subscriptions are cheap honestly. Yeah they say it's not for enterprise usage but ok whatever. Not paying $10k when $200 gets you the same value (seriously)
You have to limit yourself to each string being independent for that to be true which is a limitation that leads to the impossibility of calculating entropy generically.
You always include the measurement of things needed to run the program too.
It's a bit like how benchmarks of compression utilities should always include the size of the utility itself. Otherwise someone can just submit a program with a dictionary of 256 common benchmark files for compression and claim "it compresses them to a single byte" :)
This is under-taught in comp sci so I'll say it in all caps for visibility. THERE IS NO KNOWN GENERIC WAY TO MEASURE INFORMATION ENTROPY!!!!
Seriously. There isn't. You might say 'but measure the number of 1's vs 0's' and i'll just reply with '101010101010 repeated'. So you up it and start saying ok maybe measure frequencies of pairwise bits then and i just reply by cycling 00,01,10,11 endlessly etc. Frequency counts and any other measure you can think of doesn't actually work.
The root definition will state it's just based on the properties of the prediction but... what model do you use for the prediction? This gets to the heart of the matter;
The measure of information entropy itself is actually equivalent to creating on omniscient oracle. It requires being able to ask the question "Oh great oracle, what's the shortest program that can reproduce X"? or equivalently "What's the probability of this data assuming i always have the most perfect model to predict it?". You then calculate entropy based on that oracles answer.
Kolmorogorov, AI and information entropy are all pointing to the same thing. The reason people get confused by the fact that the frequencies of base 10 digits is the same for digits of pi and a truly random sequence is just because the counts of base 10 digits isn't a measure of entropy at all. Measuring entropy literally requires an omniscient oracle.
Meta doesn't seem to be able to produce anything close to a frontier model. The selling of compute capacity seems to be acceptance of "compute is wasted on this crappy avocado model, we'd be better off allowing something better to run".
The problem is clearly in the model architecture, the training and the data fed into the model which is causing them to give up on using their compute exclusively for their own models. They can't get it right so may as well sell the compute to someone that can.
Compression can be defined as reducing uncertainty. If you can predict the next sequence you can compress it to 0 bytes using arithmetic coding. Reliable prediction is what enables compression and it's the link between compression and AI that everyone is talking about.
No one ever in comp sci says artificial intelligence is "like compression", they correctly state that "artificial intelligence IS compression". It's absolutely known and accepted that artificial intelligence (defined as predicting outcomes with a measure of certainty and taking chosen actions towards goals using those predictions) has equivalence to compression in a very hard science way. The hardest part of artificial intelligence is compression and the remaining part, the choice of actions based on predictions is just a tree search to a goal.
It’s also amazing how bad delivery services are in general. The incentives for third party delivery services don’t align well with the other parties. A retailer is judged on the quality of delivery yet only amazon has seemed to realize this (queue incoming anecdotes about amazon screwing up delivery yet i’ve never had an issue getting a refund when it happens).
There's 'sourcing agents' whose job is to coordinate the garages. They don't work for any of the garages but as a westerner you'll get in touch with the "electric motor guy" who knows all the factories to contact for that particular purpose. They'll meet you at the airport gate and you essentially pay them as a guide to negotiate the shanghai business environment.
This is from someone that has observed Shenzen. A location where much is made in garage sized factories (usually literally a garage space at ground level where people will bang out products by hand).
You might not expect a bespoke 2 ton electric train engine to be made in a series of garages but it really is. One lot of workers will be experts at winding coils. They'll have a rig that spins and a spool of copper to wind on with a practiced skill so that they do it as well as any multi million dollar machine could. Then there will be another shop that forges an engine housing. They'll shape out a cast in sand and pour in molten steel (produced by another nearby shop) into the cast to make the housing. Another shop will make the brushes, another the motor controller, etc.
The end result? You travel to Shenzen to build a bespoke megawatt scale electric motor and you have a prototype delivered in 3 days. Not even kidding. It's not some megafactory where you will never be worth their time for an order of 10 engines to replace aging motors in a custom 20year old fleet. It's a set of people in rooms making things for low price point at exceptional scale that are easily outcompeting the western "bigger is better" style.
The USA seems crazy with it's focus on mega corps or nothing honestly. Every law seems to encourage this - eg. The healthcare system which absolutely harms small business owners who have no ability to negotiate a corporate health care plan. How do you ever develop a Shenzen style manufacturing culture in such an environment? How does a megafactory that makes a billion of one thing innovate rapidly? You need the multitude of garage workshops that collectively fill every niche that Shenzen has. Today if the West was cut off from Chinese goods we'd be stuck in so many ways. We just don't have what China's enabled here.
I think your post is a very good example of goal post moving.
We currently have debates around whether content online is human or ai generated and instead of acknowledging the milesetone we have a post essentially claiming "the turing test was passed long ago, it's worthless".
But the author acknowledged times that someone said something stupid about AI?
He's not claiming there's no flaws or that there aren't CEOs claiming more than is possible right now. He's making the point that these don't negate the parts that are genuinely impressive.
I've encountered waaaay too many "you can't possibly have done that with AI because AI occasionally hallucinates" and "CEOs say things for marketing therefore AI can't really do anything" type of posts.
TPMS (tire pressure monitoring system) is a coin battery powered computer inside each tyre of your car. They’ve been around for a couple of decades now. Even the lowest end cars have TPMS in each wheel. If you change wheels you need to go to a wheel shop and have them re pair (as in re pair wifi) the wheel with your car. I had to do this recently with my 2014 ford focus.
Anyway those are just four of hundreds of computers in your car these days.
What makes a car ‘made in China’ (therefore over 100% tariffs) vs ‘assembled in the USA’ (therefore no tariffs)?
The battery, engine and everything else is absolutely Chinese made. I don’t know how much assembly there is honestly but i feel the Geely, err i mean Polestar was a little close to that line.
I will say the laws around this indicate just how ridiculous tariffs can be. There’s always some line to press up against and honestly if electric motors, batteries, car bodies and wheels from china have different tariffs to a car as a whole it’s always going to lead to china shipping those parts in an easy to bolt together way to ‘make a car’.
No only sat scores but specifically they ask for the percentile band of your high school maths and hard sciences scores.
Not even kidding. I’ve been in a staff level+ role at 3 of the 5 faang. Applied to canonical because their products are interesting. I’m ~30 years past high school and i get hit with ‘what are your high school maths scores’. I answered the online form honestly and got a rejection email immediately on send. Phew!
Not at all kidding on that and there’s screenshots of the literally insane questions they ask online.
I think that's more about integrations/differentials not producing them (generally speaking). Physics likes to deal with integrals and differentiation as you calculate change over time or over spatial dimensions.
Eg. the integral of x^10 is x^11 / 11 + c. No hyper-operation appears and it's just another exponential (with a division).
The integral of log(x) is xlog(x) - x + c. So still basically just a logarithm
Even the integral of 2^x is just 2^x / log(2). Still basically the same thing.
There's no easy way to pull a hyper-operation out.
>The OP post is misinformation. The policy page has been unchanged since April 16 (including the words Updated this week) and has to do with verifying if you're an adult if they suspect the account is used by an under 18, which we all already know Anthropic is doing.
The other nice thing is that the batteries on cars can effectively act as grid energy storage even without v2g. Simple offpeak/low rate charging setups can take the most efficiently generated cheap power.
In Australia power prices are often negative in the day due to solar and there's various variable rate plans you can get to take advantage (Australia dwarfs all other nations in per capita solar; even China is nowhere close per capita). I know workplaces that will actively encourage you to charge your car at work.
For all the talk of 'solar can't replace fossil fuels' or 'electricity isn't green' Australia's gone and created a nation wide energy market that encourages rooftop solar and it's found itself with excess daytime energy at a time when the world has an energy crisis in Iran and the datacenters going up everywhere.
Just in general these questions are probes on curiosity and ability to show depth too. I’m astounded by suggestions of stating flat out refusal to even try out LLMs or suggestions to over praise the merits as if the interviewers want to hear binary answers. A well thought out pros and cons story wins over binary yes/no answers at pro and anti ai companies alike.
Paying by the token is insanely expensive. Only the 5̵ ̵R̵i̵c̵h̵e̵s̵t̵ ̵K̵i̵n̵g̵s̵ ̵o̵f̵ ̵E̵u̵r̵o̵p̵e̵ Biggest Tech Co's can afford that.
But the subscriptions are cheap honestly. Yeah they say it's not for enterprise usage but ok whatever. Not paying $10k when $200 gets you the same value (seriously)