150 karma · joined March 12, 2023
Contact info:
Address: 6363 Christie Ave, APT 2302, Emeryville, CA 94608
Phone: (917) 704-0699
email: hervalfreire@gmail.com
Media relations: Michelle Veronese
(https://www.linkedin.com/in/michelleveronese)
If his arguments are entirely based on this, then it's not fully correct:
- GPT style language models try to build a model of the world: https://arxiv.org/abs/2210.13382
- GPT style language models end up internally implementing a mini "neural network training algorithm" (gradient descent fine-tuning for given examples): https://arxiv.org/abs/2212.10559
To illustrate the absurdity of it consider this:
I have a theory for everything. I can predict future. The mathematical equation for that is simply (1 - EPSILON) where EPSILON is the probability of any event not happening. What about the probability of a event happening as a result of the first event? Well that's (1 - EPSILON)^2.
Since this model can model practically everything it just means everything diverges and everything fundamentally can't be controlled. It's basically just modelling entropy.
Really? No. He just tricked himself. The Key here is we need to understand WHAT EPSILON is. We don't, and Yan bringing this formula up ultimately says nothing about anything because it's too general.
Not to mention, Every token should have a different probability. You can't have the same epsilon for every token. You have zero knowledge of the probabilities of every single token therefore they cannot share the same variable, unless you KNOW for a fact the probabilities are the same.
It should be: P_n(Correct) = (1 - Epsilon_n)^n
The thing that is harder to replace is the versatility of the human form. Manual labor can't fully be replaced because robotics have yet to catch up.
>there's no need to reject the inherent value of humanity
There's no fundamental rejection here. Capitalism simply selects the most efficient methodology. If humans arent the most efficient methodology for a given task then capitalism eliminated that methodology. That's the logical extrapolation. You subjective opinions on humanities worth is irrelevant to the most likely outcome.
>Yes, this is what I mean by the limited capacity of a simple textual description. It's a fundamental limitation - natural language is just poorly suited for the detailed conceptualization.
Except LLMs can accept sketches as input. The higher order methods of communication are covered by encoders.
This is clearly different from what you are doing. You are making a hard claim that everything you see with LLMs being conscious is science fiction. Unfortunately this claim cannot be made given lack of evidence and even lack of definition of what consciousness is. Therefore by logic your statement is the one that is outlandish here.
The problem with "reverse engineering people" is that you don't need to, the things they like in any era follows predictable and generic patterns. These patterns can be encoded into AI provided we find enough curated training data.
> The hard part is to make the artists do exactly what you want, down to the subtle but meaningful details.
This is because the artist can't read your mind. It has nothing to do with your skill or the artists skill level. That feedback loop you are alluding to is more a reflection of the unclarity of your instructions or your imagination.
You thought what you imagined looked good but the artist in following your directions created something that showed you how flawed your imagination was.
The future predicted via extrapolation of a trendline is unfortunately more probable and more realistic then a future predicted via mistrust of AI.
Artists have already formed lawsuits against companies that own LLMs, this one in the article already involves an artist complaining about his job being more or less replaced.
You have to be next level delusional not to consider the extrapolation to programming.
Difficult in terms of effort. Not difficult in terms of skill. Make no mistake the quality of a movie is more the sum of the quality of the parts then it is the creative director. Who wrote the script? Who did the digital effects? Who did the lighting? Who did the editing?
The director did the hard work of picking the people and issuing orders.
>I think it's natural for humans to reduce and simplify things we don't engage with every day, or our brains would be overloaded. So we just handwave it off. We do it to other people too unfortunately... Remember how complex your life (internal and external) is; other's lives are equally complex and nuanced.
Except I am more or less a director. Not one for movies but for a company.
There is a difference between handwaving something off versus being delusional about your own role within the world. Directing is hard work, management is hard work. But none of these things are skilled work.
>All fiction writers suck? I don't know how much fiction you read, but it's incredible varied. There's a lot writers need to think about and control in the reader (and account for multiple profiles) outside of just an idea. But the ideas themselves need to be coherent. own position.
It is certainly easier to create an entire space opera in writing then it is to do it via a movie. Writing is skilled work in terms of one skill only: your ability to write. Every other aspect of it is hard work but, unfortunately, unskilled work.
I realize there are complex plots, paradoxical stories and imaginative settings and the pacing of a story is important as well. But all of this doesn't really require skill just time and deep thought to come up with. Plenty of the most popular authors never had a writing background or talent.
I would say again that these director positions while in principle are easily replaceable they are not in practice due to politics. A director or CEO is where he is mainly due to politics. Politics is unfortunately a skill with aspects that not only need to be imitated by AI but imitated by robotics as well, and it's simply a gateway into the role with no relation to the actuality of the job requirements.
Its hard to imagine a computer issuing the same exact orders as a director. But with LLMs computers are really close to doing that in principle. The issue is as I mentioned the social and political aspect of directing that cannot be replaced yet.
This is a nice way of putting it. It just shows how people have a tendency to deny a reality that may replace everything they stand for.
What you will see as AI continues to get better along the obvious trendline is these "denial" arguments become more and more specific. First the AI will "never" replace jobs, then it will "never" replace specific jobs involving "programming skills" or "skills related to what I do" until eventually when the reality of it all is too all encompassing the arguments will evolve into attacks at AI for being "low quality" or something along those lines.
But if AI progresses to the point where the quality of the output becomes undeniably superior to human output the "denial" arguments will inevitably shift to the REAL argument. The heart of it all. What is the purpose of being alive if AI is doing everything? Should we ban it for the sake of the economy for purpose?
These arguments of course only occur based off the assumption that the technology will progress so quickly that the repercussions will be hitting everyone like a freight train. If the progression slows down enough then there won't be much opposition. Only acceptance as it slowly assimilates into our society without people noticing the change.
If you in your heart are one of these people who has no worry about AI taking over jobs because AI is simply too "stupid" perhaps consider the fact that my description above fits you. Are your arguments evolving along a similar trendline? If so, consider shifting your perspective a bit.
The creative professional IS a target for replacement. What LLMs do best is specifically what the "creative professional" does. Better than coding or mathematics, LLMs excel at creative English text. This is because creative output is trivial when compared with technical skills.
I don't think the "creative professional" will get replaced though. Because usually the task is so trivial that positions like these largely only exist because of politics. You convince people around you and yourself that your "creativity" has no peer.
It amounts to this: I want a photo-realistic animation of a cowboy with blue skin with a laser cannon riding a unicycle through the grand canyon with a dwarf chasing him. <<This sentence is trivial to come up with, ANYONE can come up with thousands a variations and tweak the sentence in various ways. No special talent needed here.
A creative director simply comes up with high level instructions (like the one above) THEN the ultra hard part is making that high level instruction a reality. That is HARD and as of now it looks like both skills are being replaced by AI.
But like I said, the directors' position is in actuality largely political. Thus his position is safe even though the actuality of what he does is EASY and therefore a prime target for replacement.
How much longer will he be able to KEEP his job.
If grandma decides to purchase opioids illegally rather then get treatment that's 10 years in prison in Singapore. The law is effective in this case of making grandma pursue the correct treatment.
This is not all. Executions will occur: namely the parties responsible for the opioid epidemic will be put to death without regard to corporate shielding. Courts will determine on an individual basis who was responsible and hang them.
I believe this last point is something I wish would happen and would largely be an effective deterrent. It is entirely a matter of how much power a DA will have in pursuing such "policing" policies. The former point however is true and is the outlined procedure enforced by the government in Singapore.
Either way even without the last point I think I have shown you that "policing" can be effective. It really depends on how much policing and how severe it is.
There is the obvious side effect of possibly creating a police state by giving the police/government too much power. But while the phrase "police state" has bad connotations it is arguably not always a bad thing as exemplified by Singapore.
I don't know whether such policies will be effective in the states. But it is patently false to say that policing cannot be effective.
It is easy for a person to hold loyalty to an idealized concept that "policing" doesn't work when it comes to drugs. The reality is, like almost everything in life, much more nuanced. Ask questions. What kind of policing? How severe? Being open minded means that you need to ask those questions rather then deliberately constructing logical scaffolding to support your preconceived/popular conclusion about the US drug war.
There's not much material gain in making that grand of a claim and its actually more prone to trigger disdain (of which yourself is a shining example).
There's not much material gain in making that claim and its actually more prone to trigger disdain (of which yourself is a shining example).
The logic is convoluted, documentation is over complicated and the language contains too many primitive functions and ways to shoot yourself in the foot.
There's really nothing good that's available.
Remember what happened in Isaac Asimov's iRobot?
Additionally there's something called the replication crisis in the social sciences (psychology included) which is basically to the effect of a major discovery that most of these "black box" studies can not be reproduced. When someone runs the same experiment the results are all different.
It goes to show that either many of the studies were fraudulent or statistical methodologies are flawed or both.
Given that chatGPT therapeutic data is ALSO derived from the same training data I would say it's ok to trust chatGPT as much as you would trust psychologists. Both have a lot of bullshit with nuggets of truth.
See Japan and Singapore. Also note those two places have the lowest incidence of drug use and homelessness.
It could be because of the viciousness of the punishment. In the US the punishment matches the degree of the crime in the sense that if you sell drugs you get a punishment that you deserve.
In Singapore if you get caught selling over 500mg of weed you are executed via death penalty. The punishment does not match the crime, but rather the punishment is designed to be as effective as possible in eliminating the crime all together via deterrence and Darwinian artificial selection.
Policing DOES work, it's just most people, you included, are drug users. So you want to end homelessness, you want to stop people from over-abusing drugs, but as a casual user yourself you don't want to be put to death either or anything like that so you have to find some middle ground that works. The problem is there is no middle ground. The US drug war was middle ground and as you and your articles pointed out, it was largely NOT effective.
From a personal standpoint I don't use drugs. Morally I'm more of a middle ground type of a person. I think moderation isn't a big deal. Objectively though, since I have no skin in the game, I would support Singaporean measures to stop drugs and homelessness all together.
This is a major issue. A good build tool shouldn't require a dedicated team.
Ideally a build tool should function as sort of a settings menu for the project.
We had a googler at our company switch everything to bazel and it just slowed everyone down.
There really is something about Google engineers, this sort of subtle arrogance that the Google way is superior to everything else is existence. Maybe I'm the ass hole but I wonder if anyone can relate?
They call it theory but it looks to me more like a written music dialect.
You can't use numbers and raw logic because it has various levels of inaccuracies and things you haven't accounted for especially when the numbers and data come from times prior to the layoffs.
Reports from recruiters and people who are unemployed are telling me things that are significantly different. There is a fundamental change in the job market.
Correlation does not equal causation but you cannot deny the possibility of causal connections here.
Generally the lower tech the area than the more children you have the better off you are economically. Mainly because children can function as manual labor. With much of manual labor replaced with automation, nowadays having more children is more of a burden.
Robots have not only been about to take our jobs. They have been taking our jobs. It's just it's so subtle it's hard to see it.