Posts by Conor Stewart 62 publicly visible posts • joined 24 Dec 2021
Microsoft is just being Microsoft but this is cursors fault too. If Microsoft have in their license agreement that the extensions can only be used with VSCode then why were cursor finding workarounds and ways to hide that it was their IDE accessing the marketplace? Why are they still looking for workarounds?
They knew that what they were doing was against the license but in typical AI company fashion they don't care about licensing or copyright, they just do what they want. So yeah what Microsoft did isn't great but it was there in the license all along, it is just them enforcing it now. What cursor did however is to ignore the license and just do what they want whilst trying to hide that it was them and they are continuing to try and find ways to violate the license. I have no sympathy for cursor here.
If you read a book then someone has likely paid for that book. If you read information on a website then either it is provided for free, paid for by you or monetised through adverts.
If a LLM uses information gathered from advert monetised websites then how much money does the website make? Only the money from the initial harvesting from the LLM which will be equivalent to one visitor and that is only if the LLM loads the adverts. The website would lose out on a lot of money if everyone accesses it's information through the LLM instead.
People pay for what they consume one way or another, even if not obvious or indirectly and the owner sets the price they want for it. There is no need for a portion of profits or salaries because the information is paid for or free.
Your argument is based on incorrect assumptions.
You say you are more concerned about privacy than anything else, yet you aren't. If you were really concerned with privacy then you would be advocating for the deletion of these models and training data and nothing less. Instead you are concerned about the environmental impact and wanting the companies to not profit off of it. There have already been exploits to make LLMs output things they shouldn't, do you really think that the companies will continue patching these exploits if they have to make it public? Even if there is currently no way to extract personal information once the model is public it would be available forever giving people more than enough time to find ways to extract data or just to keep it archived until the technology required to do it is available.
Human brains are in no way comparable to LLMs. Human brains function very differently and have different functions like being able to understand and reason and have unique thoughts. LLMs are just pattern recognition. We don't even fully understand how brains work (if we did then making a copy of one would be easy) so how can you claim that brains are just LLMs?
Maybe you could do it with images or audio too, similar to how people hide data in the least significant bit of images (stenography). Maybe if that was done enough (maybe the 2 or 3 least significant bits) and the model is sensitive enough it could mess with the model.
It is punishment to the company but it is also punishment to the people who's rights are violated. It would be like if someone copied your published music or copied parts of your book and instead of being made to delete it they are just made to make it available for free. Yes it punishes the company for copying your work but it also punishes you for no reason and to most people wouldn't be an acceptable solution.
On top of that there has been research to suggest that better preparing and selecting training data can lead to smaller models and better results. An example I read about was about a model for working with images, I can't remember what exactly it could do but they generated training data by paying people to look at an image and generate a description, they would speak their description and it would be captured by speech recognition software. They got the workers to describe the image in quite a lot of detail. After training the model was smaller and better than other models that were trained on a lot more data, like the whole Internet.
One problem with the current approach to AI is that it has no understanding or reasoning, it is all just pattern recognition and advanced predictive text. It has no way to tell if the training data or it's responses make sense, this is why training data quality is so important. I think that the issue of LLMs hallucinating is because of how much nonsense data it is trained on, anyone who has spent time on the internet knows that it is full of incorrect information even amongst otherwise correct information. If an AI is being trained on this huge mix of correct and incorrect and contradictory information with no ability to understand or reason then of course it will get confused and generate nonsense or incorrect outputs.
Until AI has the ability to understand and reason it will have issues with hallucinating but this can be partially avoided by being very careful what you train it on. If AI could understand and reason then it could weed out the incorrect information and check what it is outputting makes sense and is correct. However we are a long way away from AI being able to understand or reason, it is a massive jump from current "AI" technology and what we have now is basically decades old ideas (like neural networks) taken to the extreme.
All universities have different arrangements. The one I went to states that students retain ownership of all their coursework and research because they are not employees so their IP does not automatically become the universities.
For employees it is different, the university owns the IP if it was created during their normal duties or if it was created using university resources. Research is either subject to those rules or can have its own arrangements. Externally funded research is different still and has individual arrangements between the researcher, company and university.
Even in the cases where the university owns the IP the inventor still has a part in it, they still receive a portion of the profits and still have some rights. The same applies to spin off companies, they need to pay the university a portion of their income.
If any third party can just copy someone else's research then companies will just stop researching. Say a company spends millions researching a drug and then after all that research another company just comes along and makes it themselves and sells it at half the price, what do you think happens to the company that developed the drug? They can't compete, they make no profit, they spent lots of money on development so the company either or goes bankrupt or stops researching, either way the research stops. Why then would companies research if they know they will just get copied and undercut immediately and the whole process will just lose them money? This doesn't lead to society improving long term. It may be good for some people short term but long term it just means no more research.
Your comment about extending IP protection to AI generated IP is just not a good idea in any way. LLMs are trained on data gathered without permission and everything they generate is derivative of its training data, there is nothing truly new or unique generated.
I am not the person you replied to but whilst we do need a patent and IP protection system it needs to change. In a lot of cases now it is abused and used to stifle innovation, both with companies patent trolling and patents being too long and too broad.
In my opinion patents need to exist but need to not cover entire concepts and abstract ideas, instead they should only cover implementations and only if the company actually uses the patent. For technology now 20+ year patents on whole concepts is too much.
As an example for pharmaceutical companies, a patent on a specific drug is absolutely fine, even long term, it is a specific implementation and other companies would only copy it, not innovate on it. However patenting a whole class of drugs or concept is different and should not be allowed in my opinion because even though the company may make drugs within that class of drugs it slows down innovation and possibly prevents good treatments from being found sooner.
Another example is 3D printing. The entire industry was held back due to 20+ year patents by stratasys on entire concepts.
In my opinion patents should be used to prevent people from just copying implementations, not for preventing innovation.
Pretty much, just like they were with crypto mining. They had no direct hand in crypto mining they just built and sold hardware that could be used for crypto mining. They made a huge profit that didn't come from crypto mining themselves. So Nvidia is very much like equipment manufacturers in the gold rush.
It reminds me of parking tickets. If you are rich enough then you just don't care, paying the tickets is not a problem and they wouldn't even notice the difference in their bank account, they don't even pay them themselves, they just tell their assistant to pay them. They know they will get a parking ticket but there are no real consequences so they don't care.
Fines for large companies are just the same, the fines make no financial impact or have any consequences so the companies do not care. Or they use their expensive legal team to get out of it.
Just because it benefits you that doesn't mean it should be allowed or legal.
It is theft, plain and simple, if you applied your logic to anything else you probably wouldn't agree with it then. Is stealing money from people fine if it is given to charity or to poor people? Is stealing houses fine if they are given to poor people? If I buy one copy of a book and then print many copies of it myself and give them out is that fine? No none of that is fine, it is all theft.
That code you are writing with speech recognition, are you actually speaking every word and piece of punctuation yourself or are you basically telling the LLM how you want other people's code stitched together?
Everything you described can be done legally without using people's stuff without permission. Yes it won't be as easy and will be more expensive but it is possible.
Weather forecast models shouldn't have any problem with finding legal training data, neither should medical models, neither should financial models. Your statement that, "all the systems are based on a mixture of public domain and copyrighted information", has absolutely nothing to back it up. Also the medical industry has a requirement for privacy and laws specifically made to deal with medical privacy, at least where I am if anyone outside of who is actually treating me wants access to any of my medical records for any reason then they have to ask permission.
It has also been found through research that if you prepare the training data better and are more selective with it then it is possible to train a better or just as good but smaller models on less training data. The data can be prepared better by manual detailed labelling as one example that can be used for images.
For good speech recognition then why do you need to use data without permission? Either hire people to create training data or ask for volunteers and market it as helping disabled people. Ask people to spend a little time, maybe even just 5 minutes, reading out some text to be used as training data. This is something that people like you could do for yourself since if it is just your hands that don't work well then you can still read and speak. Yes it would be more work than just stealing every bit of audio with a transcript that you could find online but it is the morally and ethically better way of doing it. There would also be the requirement to use the data for a specific purpose and nothing else, like if it is collected for speech recognition then it can't be used for text to speech without asking for permission again. If it is code that is collected for a code speech recognition model then it is only used for that, not to train a code generator model.
There is no excuse for stealing, even if you believe it is being used for good. There are ways to get training data in a moral and ethical way, but these companies don't because that is more work and would cost more money.
It doesn't matter how significant the data is or is not, it is illegal to use it like that. Also if a lot of the data is used without permission and illegally then it should all be removed, so even if an individuals data is insignificant all of the illegally used data combined is definitely not insignificant. You talk like it should be an opt out system after the fact, why? Why shouldn't it be an opt in system like it morally should be?
Since you seem to have all the answers. How do you remove training data after the model has been trained? It's not a simple task and likely isn't even possible due to how the models are trained and would involve removing the data from the training data and retraining the model which is not feasible to do every time someone wants their data removed considering the cost and time required to train the model.
Plagiarism is deliberate copying or stealing of ideas, it has nothing to do with how unique a concept is. If someone's IP is not unique that doesn't mean they plagiarised. There are many examples of the same technology or ideas being developed independently at the same time.
"The GCSE contained out-of-date content about networks and internet protocols that could be removed from the specification"
What out of date content would this be? The Internet and other networks still work very similarly to how they did when they were invented, most of the content should still be relevant especially content taught at GCSE level.
Or is this out of date just because most people don't need to know this? Instead they just use a library and it handles everything?
They might well have good learning material on how AI works but that isn't the point. They want a separate course that is just about how to use computers, not computer science and there is a very good chance that it wouldn't go into how AI works just how to use things like chatGPT. So at best it would be surface level knowledge on how to use particular models.
'EV motors are “simpler,” and manufacturers don’t need a “decade of experience to make a good electric motor,”'
Spoken like someone who doesn't know what they are talking about. Electric motors are far from simple and don't exist in a vacuum either, they need a lot that goes along with them.
Sure you could make a basic DC motor and run it with simple drive electronics but you wouldn't get good results. Motors are very complex when you start trying to optimise for power and efficiency, especially when you also try to make them light, compact and reliable. Then add in the battery and it's protection and monitoring and charging systems, where they are trying to pack as much capacity in as they can and still keep it small and light, whilst still being able to provide high current. Then add in the very complicated area of motor drivers and regenerative braking, designing a motor driver to handle that much power and do it efficiently and handle regenerative braking is far from simple.
Then add in all the other parts of a car, air bags, crumple zones, brakes and all the electronics and that is far from simple too and it needs to be very reliable and fail safely and pass all kinds of regulations. Why do you think most EV manufacturers are ICE car manufacturers?
I have no doubt that he would have a lot to say if someone called core parts of the linux kernel simple and said that they don't require much experience to make. I would have thought that with him being such an experienced developer and managing a large scale open source project that he would realise that things aren't that simple. It is a common annoyance with engineers and software developers when other people who don't know what they are talking about call something simple or easy when it really isn't.
For some reason I have seen that a lot of people seem to think that you can use a Tesla or other EVs 12 V battery to jump start an ICE and have heard of people actually doing it. Whilst it may work the batteries aren't designed for it and it will just degrade them and the manufacturers don't recommend it either
It is probably a safety thing. By having a small 12 V battery they can completely disconnect the main battery (from everything except the BMS) when not needed. It also means they don't need to deal with the inefficiencies of a multiple hundred volt to 12 V converter when the car is just sat there since the quiescent current would just end up draining the main battery.
But knows very little about car design. You would think someone with his experience working on a large scale project with lots of different interoperating systems he would realise that the engine/motor of a car isn't the only thing needed and that an electric motor, battery and motor driver is actually quite complicated. I bet he would have a lot to say if someone refered to certain parts of the linux kernel as simple and could be done without much experience.
Whilst ICE are not very efficient, EVs aren't all that efficient either once you take into account all the steps and losses, from generation to transmission to the EV charger to the batteries and finally to the motors. Most sources I saw claimed an EV itself was around 70-80 % efficient which when stacked will all the other inefficiencies. Obviously fuel engines have similar transmission and extraction losses too though.
Due to the nature of EVs there is only so much they can do to make them efficient since they need to be light and compact.
Then there is the issue of batteries, particularly lithium ones, they just aren't good for the environment due to how they are produced. Batteries are what hold most electric vehicles back, especially aircraft, they are just nowhere near as power dense as fuels. Hydrogen is likely a better choice long term. Faster to refill and doesn't require large amounts of lithium batteries, plus it is clean too, the only issue is storage.
It was such a dumb comment about something he clearly knows very little about.
Your comment about regenerative braking misses something out. Nothing is 100% efficient and so if you accelerate with a full battery up to 60 mph and then regeneratively brake back to 0 you won't end up with a full battery. The only time what you mentioned would apply would be if you started fully charged on a hill going downwards. There are other ways of braking using the motor too, if I'm stead of regeneratively braking you instead short the motor windings together then the motor will brake too. Also all EVs require friction brakes anyway for emergency stops and incase the motor or motor driver fails and in case the regenerative braking isn't aggressive enough.
For the most part I agree with you, but I do still believe it could be used badly. Even just a project being taken in a direction that isn't beneficial for most people or in the original purpose of the project just because a company is giving the developer money. At what point does it just become writing code for the company?
It all depends on the integrity, stability and judgement of the developer. I could see it happen quite easily, a company threatens to pull funding that the developer needs to continue or sustain themselves if the developer doesn't do what the company wants.
I can also see it turning into a competition between companies. They each try to donate more than the others to try and get priority on their requests. If could be very good money for the developer but it may not be worth it.
Yes developers need to eat, so they should get jobs or charge for their code rather than just release it for free. They could also offer additional services like support or hosting. If their project is large enough then maybe they can sustain themselves on donations but that only happens with the largest projects and it shouldn't be expected for every project or piece of open source code.
Why do people release code as open source? It shouldn't be so that they can make money on it.
Relying on donations for your open source project to sustain yourself is not a great idea anyway. What if someone else releases something better or even just forks yours and improves it? What happens if a company releases closed source software that is much better? Realistically your project could become obsolete very quickly and then the donations will drastically decline.
As far as I know it is only modifications to the open source code that needs to be open sourced, any other code doesn't. So if a company uses some GPL licensed code in their software they only need to release any modifications to the GPL code, not their whole code.
"Corporations should not sell any software based on the work of Open Source GPL software. Period.", why? Because you say so? If you want to stop something like that then put it in the license but as it is just now companies are allowed and able to use GPL licensed code
OpenAI is particularly bad. Yes they started as being open but for a long time now they haven't been but won't change their company name, likely due to the public perception of it having "open" in it.
There are no protections for this sort of name though and that has meant a few companies have done similar, start off being at least partly open with "open" in the name and then go closed source.
Balancing absolutely is still needed and your idea of slow balance charging isn't good. What do you mean by "low power resistors" being used for balance charging anyway? Balancing is even more necessary on high current draw devices like power tools and again necessary on battery packs that are abused, like most power tools are.
By going with your idea then a pack that is even mildly out of balance every time it is discharged will then have a lower capacity unless you leave it to trickle charge. If you want the full capacity you have to wait for it to slowly balance itself when you could just use a better charger or balance circuit that does it properly.
From being around tradespeople a very common complaint is that batteries and tools from different companies aren't compatible. Never once have I heard them complaining about the battery chargers, let alone wanting them to be usb C. Most of them wouldn't want to have to use a phone charger and USB C cable, especially not because of the fragility compared to the current method of charging power tool batteries. It also isn't all that difficult for them to keep the charger with the power tool and all the other stuff they need to take around with them.
This complaint/idea of yours seems to come from someone who only uses tools occasionally and casually. Regardless of any kind of tough USB C connector or standard, the current methods of connecting and charging these batteries is much more robust without small pins and contacts to get filled with dirt, dust or sawdust or damaged and is physically much more resistant to impacts.
You are trying to force a connector into an application that doesn't make sense.
PPS only supports up to 21 V, that is just enough for a 5S lipo or li-ion battery. It wouldn't work for anything higher.
Also for anything more than a single cell you need balance charging which is sometimes done in the battery pack but is often done in the charger. So the PPS functions are only useful up to a point, you would still need extra circuitry and converters in the battery pack to be able to balance charge.
The classic argument, it's the universities fault. They teach the kids and that is what dictates absolutely everything to do with the whole industry apparently.
It's a old argument and one that doesn't make a lot of sense most of the time it is used. If it was all just left up to recent university graduates and what they learned at university then what do you think would happen? It wouldn't be what we see happening now.
Nvidia are doing pretty well in the AI and general accelerator/GPU market so I would think it pretty unlikely that they are licensing sifives designs for that. If Nvidia switched over to using SiFives architecture instead then they would likely lose out on their walled garden they have created with CUDA. I can't see Nvidia licensing a potential competitors design when they have already cornered the market.
Maybe Nvidia is interested in the RISC-V cores for their Jetson or Tegra line or similar rather than using Arm cores?
The parts they are talking about existed long before 3D printers and can be made by other methods.
They say they are smaller than a USB stick and can be made by 3D printing? Then they can likely be made by someone with a block of plastic and a file then and plenty of other methods.
The only reason they are seeing an increase in the number of arrests for them is because they are looking for them now.
That is part of the argument I just don't get, how can they genuinely say that people won't need to learn rust? If rust is part of the Linux kernel and I want to work on the Linux kernel then logically I need to know rust. If I don't know rust then what happens when I come across a section written in rust that I need to alter? Then I have two options, don't alter it or learn rust.
Sure when rust is just used for device drivers and components then it may not be essential to know just now but they seem to want it to become a major part of the kernel, so how would you be able to get away with not knowing it then?
RISC-V is praised for being an open source ISA so anyone can implement a processor based on it. If professional processor designers, intel, AMD, T head, etc can get it so wrong at times then what hope do the rest of us have?
People often talk about how a company can create their own RISC-V processor for their product that is exactly as they need it, especially startups so that they don't need to pay licensing fees and how we will now have competitors to ARM, but is this really what we want? Do we want our devices to contain processors designed by less experienced teams with likely less rigorous testing?
My trust may be misplaced but I would rather a company use an ARM core (or a reputable RISC-V alternative) in a product than try to roll their own RISC-V core, exactly because of issues like this which will likely only get more prevalent as more people start designing their own cores, I really don't see open source cores as a good alternative either unless they have been properly tested.
He said they were never able to pin down which regulation he was breaking so it seems either he didn't know about the regulation or there wasn't one.
Is there any regulation that you can't drive after a 9 hour shift? If so then loads of people are breaking it.
It's probably more likely that the officer saw he looked tired and decided he wasn't fit to drive.
That is all they are, fancy predictive text, it has no understanding of what any of it actually means, it is all just probability based.
People either don't realise this or they come up with excuses, like saying you need to prime the model first and lead it to the answer, but if I go to a human expert with a question, I just ask the question, no need to prime them and they will tell you if they don't know.
During the trend of YouTube videos using chatGPT to write code for arduinos, the code it came up with I was sure I had seen before, it was near identical if not identical to the code on tutorial websites. People were praising chatGPT and saying it makes it much easier to find information and it can write code for you but you would have better results just searching for what you want and clicking the top link. ChatGPT was just regurgitating it's training data and then throwing together some explanation. For all intents and purposes chatGPT was acting as a search engine without crediting the source material.
Exactly, why is it an issue with Meta running imprecise calculations for their "AI", which by nature doesn't need high precision, but not any other application requiring precise data? Why isn't this a huge issue in banking, stock trading and weather simulations?
Either Meta is using sub standard hardware with no error correction and detection or it is all just an excuse.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
