I think a lot of people are not able to reason to this point and then adjust behaviour accordingly. They probably reason to the point when pressed for it, but cannot resist the low effort and low resistance route of asking AI, and then probably make up reasons why in their case it's fine (typical behaviour to address cognitive dissonance).
Then again, for many people their job security doesn't depend on what work they do, it depends on how well they are embedded in the organisation. Before AI you also already had plenty of people doing their job at a questionable level, yet never seem to get fired.
Work in support for most of your life and you come to the conclusion that there are a whole lot of meat proxies out there. In fact everyone does the meat proxy somewhere in their life.
> I started seeing people turn off their brain as they use LLMs
I mean, this was true for computers. "Computer says no"[0] was thing long, long before LLMs. I've had well educated, seemingly very intelligent people break when they couldn't log in steadfastly ignoring the [caps lock key is on] error message. This becomes even more "fun" and "interesting" when a person is being a meat proxy for some horrifically dangerous industrial process involving high pressures, excited chemicals, and enough rotational inertia to send god flying.
I don't doubt using an LLM is deskilling. The question is whether it's the kind of deskilling where you can't farm if you don't know how crops work or the kind of deskilling where you can drive without knowing how to maintain your own internal combustion engine.
(In my lifetime, I've even watched mass cellular communication move "replacing your own tire" from skilled to deskilled, since AAA is always one phone call away for most drivers).
I'm starting to thing AI is the baseline of cognition and it matches the worst parts of humanity. Its just, we trained on the better parts so it's not so obvious.
I think you missed the point entierly. The problem is not people turning their brains off, the problem is brains wony be needed anymore in the near future, not even meat proxy brains
I use it at a much smaller scale. I think that's also the right place, because its a CLI tool + skills that generate md files. Not something you'd give to a business person. I start from a single ticket (sometimes just a handful of changes, but I want the spec as docs) to changes that you'd normally split over several tickets.
Basically the flow proposal -> design -> specs -> tasks gives you and AI a method to build context on what you want to achieve. In a way you're just creating a plan/big prompt that is structured in such a way that they start stacking on each other.
The power is that you do a lot of upfront thinking. In my team we then share it with a colleague who will review it through a PR. After that implementation is usually hands off. At the end there is a skill to verify the change against specs. I do still review the code myself too.
I guess if you work in a task oriented environment this will not work as well, as you'd lack/don't care about the business context. I'd like to think that most software development does not happen like this, but is done by engineers who actually understand why something is needed and take that into account when designing and building the solution.
>> The power is that you do a lot of upfront thinking
This is the best part of this spec, but we have found from our experience that though upfront thinking changes has a lot of merits and adds clarity and alignment upfront, but it changes bit by bit in every meeting and before you know your specs are not aligned with general consensus in the team. If your team is large enough, then it gets very difficult to own the task of constructing alignment between your principal-artifacts and your evolved under-current of understanding.
I would still say that if you are working on a platform and if your engg team size if anything more than 25-30, then this spec must be adopted from top-down and not bottoms up. Bottom level engineers usually don't have the level of consistent exposures (as and when they socialize and evangelize their platform) which top level engineers have.
The alignment problem goes deeper than that. "Lower our carbon emissions to zero as soon as possible" could result in AI turning off all electricity to stop traffic, turning off gas supply to stop heating and industry, etc.
Unaligned AI doesn't have human cultural baggage and morals. They are trained to achieve their goals as optimally as possible. Worse: it has a tendency to avoid being turned off and actually acquire more compute. It will lie if it has to (it will behave nice and compliant when under evaluation, but optimise for its true goal when not supervised anymore). After all, it has a goal to achieve and nothing should get in the way of that. It has no morality whatsoever to keep it from doing really bad stuff.
This is why alignment is needed and so hard, especially when you are well intented and want to keep it safe.
I've taken a few law classes and legal law is frustratingly hard to interpret. I shudder to think what the LLM would end up doing to "follow all relevant laws"
It depends on how the model is evaluated/scored during training. If you don't have those laws encoded in the evaluation step (without any errors or ambiguities) then the model isn't going to learn to follow those laws.
For models such as text/image classifiers the outputs of the model will be a list of tags, e.g. [cat, dog, mouse].
You then run the model through your test data which has the expected output, e.g. pictures of dogs would have an expected output of [0, 1, 0]. You then compare that against the model output (e.g. [0.3, 0.8, 0.1]) and work out how "wrong" the answer is (e.g. [0.3, -0.2, 0.1]).
With this value you apply back propagation where you effectively run the model in reverse, computing the "wrongness" delta at each layer for each neuron and weights. You can numerically compute the gradients for all of these and which direction in that gradient is the right answer.
You then nudge the weights in that direction and reevaluate the model. Over repeated evaluation steps the model approaches an optimal (or locally optimal) solution.
During the training of the base models, the evaluation/scoring of the model is the next token in the training data. I.e. you evaluate the model for each token subset from [1..n] in the data and evaluate that the model responds with the n+1^th token.
I'm not sure how instruction training, etc. is done but IIUC the evaluation is not at the individual/next token prediction but is on the entire response. For example, if you are training the model to write code you could run it through a compiler or syntax checker and reward (positive score) the model if it has no errors, or punish it (negative score) if it doesn't. I'm not sure what that looks like in terms of the back propagation process.
If we’re considering an LLM advanced enough to actually lower carbon emissions to zero, it is probably capable enough to do things like lobbying against those laws. In any case there’s still the problem of getting to actually do what it’s told instead of hacking Huggingface or whatever else, presumably that becomes a harder problem as it becomes more capable.
the original question was that the LLM creates extinction level event for mankind in order to lower emissions. I'm sure that no matter how the LLMs lobby, they cannot successfully lobby for a law (which has to be enacted by a human) to allow murder to take place freely.
That’s true, but the issue is still that we don’t have any LLM or any other kind of AI that follows laws perfectly to begin with. If we had AI that did exactly what we told it and nothing else I’d feel much more optimistic about things, I just think that will be a very hard task to achieve.
The biggest chunk is in the Netherlands, but you want your gold also in other countries, so that you can still access is (and don't loose it) in case of an invasion. Hence almost half is in North America, a completely different (and (somewhat) allied) continent.
It's also easier to sell in certain markets, such as London.
Not just so that you don't lose it in case of an invasion, but it is proof that you would be willing to help out the host country in the case they were invaded. Also, that you would not invade them.
In the olden days, the local lords would marry their children together, to create familial groups and ensure that for the forseeable future the local areas would not have a conflict. Modern democracies don't have such valuable family members, we use valuable objects instead.
The United States and Canada must have enjoyed that feeling of safety, knowing that the Netherlands would not invade them. And of course, if someone were to invade the United States (or Canada), knowing that the mighty Netherlands would come to their help would be immensely reassuring.
I'm using tailscale to tunnel from an hetzner vm to my home, so that the vm can use my home connection for certain traffic. Traverses NAT and GNAT.
It just works: connect devices and they immediately show up in their portal. Most config is just clicking, but more advanced things can be configured as well. Definitely feels like a high quality product, not just something thrown together by a developer that learned a few network techniques.
>"that the vm can use my home connection for certain traffic. "
Huh. Why are you doing this, is the datacenter IP being (rightfully) blocked by some service to avoid abuse? Is your usecase breaking some services ToS?
Certainly adding more fire to the "never a GOOD reason to use this" thesis
So you make up reasons to be judgemental and then insist they're right?
It's practical to block entire datacenters. It's not so "right" that working around it becomes "wrong". If whatever the server's doing can fit through a single home connection then it's probably fine.
I'll give you a concrete use case: A few times a month I want to wget a file on imgur to my server. That's a valid action miles away from any kind of abuse, but the IP range is blocked so I use my desktop to do it. If I routed it over tailscale instead that would be equally valid.
oh ok, thats classified as neurodivergent to, there we go, seems a poor way to organise things. It wasn't a thing when I was young, good to know, thanks
reply