s is defined in the statement and then is undefined out of that statement because there is no other scope for it - it's at top level. just write s =0 and all is well as s exist in the global scope where your code lives as well...
7,940 karma · joined January 24, 2012
s is defined in the statement and then is undefined out of that statement because there is no other scope for it - it's at top level. just write s =0 and all is well as s exist in the global scope where your code lives as well...
Can you really describe a process like EUV lithography an optimisation? I mean it requires matter to be controlled and manipulated in a way that would have been regarded as pure science fiction 20 years ago. Also the material science that provided Gallium Nitride electronics in our communcation system is rather amazing. There are other things as well - I have an electric car with a battery that lets me travel for hundreds of kilometers, if I trusted it, it could take me there without me operating it (much). I know where I am in London and where I am going because of satellites in geosynchronous orbits and calculations that use relativity. Last year I got an mRNA vaccine, that's pretty new... and pretty pervasive. I've seen rockets that fly up into space and then turn around and come back down to land, and I spend my days talking face to face with people on the other side of the world. I've never met half of them, but they greet me as an old friend.
How is it that you can't see any of these wonders that have sprung into being in our life times?
Folks who do this should be shunned by their peers and driven out of academia. It's exploitative and coercive.
Unless closed models have significant advantage AI inference will be a commodity business - like server hosting.
I'm not sure that closed models will maintain an advantage.
I have actually worked in academia, and in industrial research. I've served on many program committees and participated in peer review actively for many years.
I am sorry to say, I am one of the people who should be asked.
The system is corrupt, coercive, exploitative, and delivers poor results.
Things were much better when academics wrote 20 papers in a career.
https://blogs.lse.ac.uk/impactofsocialsciences/2025/02/17/do...
I am afraid that there is massive and systematic and harmful fraud in research: https://www.science.org/content/article/potential-fabricatio... as only one example. There are many many others. Very little work in entire fields is actually valid.
I think that the way that authorship is currently practiced has a lot to do with this fraud. Academics with ethics and goodwill can no longer effectively police the system. A large number of academics do not even want the system to be policed.
https://blogs.lse.ac.uk/impactofsocialsciences/2025/02/17/do...
I am afraid that there is, likely, a correlation between publication rate and fraud. I agree that a very high publication rate doesn't necessarily mean fraud, but I am afraid that it does cast suspicion, in my mind, on the totality of the output of the author.
In my field I do know some legitimately high output authors. I know a lot of authors who think that they are legitimately high output, when in fact they are simply gaming the system. The sad thing is that they don't know better. I know a lot of people who believe that they have no option but to go with the flow as well - but know perfectly well that they are acting badly.
This is not just a question of academic morals. There are children who today will receive medicines that have no value, and may harm them, because of this practice. There are lines of research that will lead no where and produce no value that are being funded because of this practice. There are lines of research that would provide significant societal benefit that are not being funded because of this practice.
Now we hear of academics (with their names on the papers) denying that they had anything to do with the work published, and at the same time the total fraud of academics with their names on 20+ papers a year.
Something has to give - and I think we all know what....
Can you link - I am answering an email on monday where this info would be very useful!
It's been very important to the Nato stance to have multiple nuclear armed states, it made Soviet calculations extremely difficult. With the USA leaving the alliance I think that either a joint Euro bomb needs to be developed or Germany needs to acquire one. Perhaps a polite fiction (continuing the current one by the USA) of France providing nukes for German, Dutch, Danish and Polish war planes could be used.
Yes: if Germany acts as a serious and honest partner in defence.
No: if Germany refuses to be a responsible security player and swings back to Russia as soon as it can. If Germany reopens the gas pipelines and throws the balts to the wolves I think that they may find that when they need help, there will be no help.
Also the UK and France have very limited nuclear capability. Yes, it's really helpful in the context of the Nato alliance. The uncertainty that the european weapons adds into any calculus of escalation really enhances the overall deterrent. Once the USA is out of the picture there is no escalation ladder, it's national extinction (and)or bust. Very unstable, untenable. I see no evidence at all of this position being altered or updated. If Germany could depend on France or the UK in a nuclear blackmail scenario then I would expect (at the least) a crash tactical nuclear weapon program to be up and running in either one or both. I also think that a four vessel continuous deterrence capability is inadequate, so I would expect much more co-ordination between France and the UK on the at sea element and/or a shift to five+ vessels.
I see none of this - the opposite in fact.
Genuinely - this is a one way bet for the Russians, smash up Estonia and extort the Germans, spend the money on more guns and then kill all the Poles.
Coming to a continent near you soon!
There are two levels...
The pZombie type level where we look at the LLM as if it were a black box and simply account for its behavior. At this level LLM's claim to have knowledge and also claim knowledge of their limited knowledge "I can't actually taste". So approached from this direction we are in (2) they have awareness that there are some things that they don't know, but this awareness doesn't prevent them from pretending to this knowledge.
If we consider it from the perspective of knowing what's happening inside LLM's then I think the picture is different. The LLM is doing next word prediction with constant compute time per token - the algorithm is quite clear. We know this is true because it runs on llama.cpp or mlx on our macbooks as well on the farms of B200's that we fear will destroy the atmosphere. So LLM's don't have any actual operational knowledge of the logic of their utterances (dunning kruger, dunning kruger...) What I mean is that the LLM can't/isn't analysing what it says, it's just responding to stimulus. Humans do do this as well - it's easy to just chatter away to other people like a canary, but humans also can analysis what they are saying and strategically manipulate the messages that they create. So I would say that LLMs cannot be concerned about what they do or don't know - the concern rests with us when we challenge them (or not) by asking "how can you know that chocolate tastes better than strawberry - you have never tasted either".
"not on board", "anti-innovation", "not a team player", "disruptive", "unhelpful", "negative".
bye bye bye bye....
I see a lot of devs and IC's taking the attitude that "facts are facts" and then getting shocked by a) other people manipulating information to get their way and b) being fired for stating facts that are contrary to received wisdom without any regards to politics.
Blackfaces can live all year on the hill, if you get clothes made from their wool, so can you :)
So the good message here is likely to miss the mark where it may be most needed.
- high wages are not evenly spread demographically in the USA
- the wage distribution seems to me to be unusual in that it has a very long fat tail whereas the UK's is very clustered (I've forgotten the right term) on the median.
- modern industry can be highly automated
- modern logistics mean that industry can be decentralised
I believe that the last two are new since the USA and Europe outsourced large amounts of their industry to China. However the bigger issue is that competitive industries require very significant capitalisation because on the one hand modern products are staggeringly well engineered (with the trade offs of cheap, good and sophisticated taken into account) and on the other hand the processes used to make them require lots of tools, infrastructure, and robots.
I think that the onshore impacts will be much lower or negligible, or possibly even positive, because so much work has been offshored already, and as is well worn in every discussion, Jevons paradox may drive up demand significantly (to be fair I believe this as wherever I have worked we've had 3x+ demand (with business cases) for development projects and had to arbitrarily cull 2x of it at the beginning of each year. So, just like the 30 people in India that are working on my project won't do anything useful unless we feed the work to them, the LLM's won't do anything useful either. And just like we have to send lots of work back to India because it's not right, the same is true of LLM's. The difference is that I won't spend 4 hrs on a friday afternoon on Teams discussing it.
But this is not surprising because we've had big impacts from tools like IDE's, VM's, and compilers which have driven seismic changes in our profession, I think that LLM's are just another one of those.
What I'm watching for is an impact in a non tech domain like healthcare or social care. These are important domains that are overwhelmed with demand and riddled with makework, yet so far LLM's have made very little impact. At least, I am not seeing health insurance rates being cut, hospital waiting lists fall or money and staff being redeployed from back office functions to front line functions.
Why hasn't this started?
For this problem we don't care if it's possible that sometimes there are things that aren't verifiable, or the answers aren't exact, we just need training signal.