Where do we go from here? (Regarding AI)
TW: AI discussion.
Recently, things have been getting rather crazy with AI to the point that I think many are being forced to reckon with its potential. It has been demonstrated that this technology is capable of solving things like millennium math problems, which has drawn response from many fields medalists. There have been very public instances of big tech losing control of their experiments. Many top researchers are expressing discomfort. Now, of course you could dismiss this all as marketing but I personally think that would be a bit of wishful thinking.
I would like to have an open discussion about the current moment in time. Therefore, I ask you, people of tildes, where do you think we are and where do we go from here? Please feel free to pull out your crystal ball and speculate wildly. Or, please feel free to call it all bunk. I just want to hear your opinion.
- Do you feel concerned about AI or are you still unconvinced?
- Do you think AI will take your job? Why or why not?
- Are you considering a career change or any other big life decisions because of the changing world?
- How are you advising your children, regarding schooling and careers?
- Do you have a subscription to an AI service? If you have one, how do you feel about it?
- Do you think that this technology will be a net positive or a net negative?
- How do you think the human race will change, if at all?
- Do you think we will see a serious incident driven by AI in the near future?
- Does it bother you that humans may not be the dominant intelligence at some point? Do you think this is even possible?
- Does any of this affect your sense of identity or individuality? How do you feel about your skill set and hobbies?
- What do you think it means to be human in a machine dominated world?
- Overall, are you feeling optimistic or pessimistic?
For me personally, I have shared my opinion many times before. It is a rather pessimistic one.
I recognize that this is a heated topic, and I would like to request that even if you disagree with someone, that we treat each other with respect and kindness. Thanks.
I am more angry than worried so far. Mostly at politicians and lawmakers to be so slow to respond to how these big tech firms just go rampant without any responsibility. If your AI starts hacking other systems, then you need to go to prison. It is not an excuse that the AI systems are too complex to understand. It is still software at the end of the day, and the people/companies running it still have full responsibility. But as usual, the tech sector has gotten away with doing basically everything they want without regulation, so we will probably have to wait a while before consequences hit the right people.
I work in software development and AI at work is of course a thing. I am just not seeing the massive revolution yet so far. It boosts things in one place, but also creates a myriad of new problems that takes time, and mainly just moves the bottleneck to other steps in the big complex process that is the eternal continuous work on software systems with decades of legacy. In short, I find AI tools to be pretty helpful in starting things, but not at finishing. The 80/20 rule of software development still applies (the last 20% takes 80% of the time).
For society in general, outside industry specific good use cases, so far I see it as a net negative to society. The amount of noise, slop, scams, misinformation, erosion of peoples ability to think for themselves, amount of money piled into it, far outweighs the genuinely useful uses of the technology. Just an example of small things, like now birthday party invitations or yardsale flyers are now ugly AI generated. It is just doing a poorer job of what we didn't had big problems doing ourselves before. AI is being used a whole lot to do basically just make a poorer crappier version of what we could do ourselves before. Now just using more energy. That is hardly progress in my book. Maybe that will change, but I am mostly just dumbfounded what so called average people are using AI for. It is very minor things that barely needed a solution.
I personally really don't agree with putting this on politicians/lawmakers and them being slow. The pace of change is just so very fast it is hard even as an individual to react. Given that politics is in general fairly slow process, expecting well thought through solutions from that end is too much to ask. I might be a bit biased as I am semi-politically active, I just find the "blame the politicians" to be a bit ignorant. Politicians aren't some special breed of people that sit and wring their hands coming up with stupid ideas. They mostly have some ideological basis for their proposals, which can be fairly understandable if one tries to see things through that lens... Anyway, I think people should mostly just join some political party instead of being mad at "those politicians over there".
As for the hacking bit. I agree. That's a law enforcement issue though...
This will likely pass (I hope). But, I agree it's both annoying and unnecessary.
Edit:
This is where LLMs struggle the most. Give it a greenfield project and you get to a legacy system much faster though ;) Improvement!
Respectfully, I agree that this space moves too fast, but to me that should mean politicians should be more proactive rather than reactive towards this technology.
I watched OpenAI’s presentation about their hacking of Huggingface, and it was basically an “Instrumental Convergence” in action - you might be more familiar with the “Paperclip maximizer” thought experiment, where an AI exterminates the human race in its pursuit of maximizing the production of paperclips.
(A very rough and simplistic version of the hack: the agents were trying to get access to the internet to get some answers to the problems they were meant to solve - huggingface just happened to be one of the ways they figured how they could get there)
I don’t expect politicians to be able to understand it fully, or be able to predict how these things will affect their nation, but I also don’t think they should just sit tight and hope for the best.
Sometimes I feel like some people (not to signal out OP, but I've seen this sentiment shared elsewhere) live in a parallel universe to me. I am not concerned by AI developments in the slightest and never have been. It has no bearing on my personal life and I have no worries about my professional life being negatively affected by AI. I use it for work, in which I get generally poor results from Claude Opus.
It has had no impact on my hobbies or plans in life, let alone my sense of identity or my understanding of where humans stand in the world.
If it was able to solve a Millennium Prize Problem, I will sleep the same at night (although it looks like that's currently in a heated debate and there's decent odds it just finished off existing human work).
I am concerned more about the public reaction than the tech itself. It's depressing to see how many luddites there are. People no longer trust science and technology.
To be clear, Luddit is a someone who "protested the usage of certain types of automated machinery owing to their concerns over worker pay, labour savings, child labour, working conditions, and output quality." (from wiki).
It was never against technology per se, it was always against corporation culture. As far as I know.
And it’s the perfect analogy, whether either of you meant it to be or not! A movement triggered by the economic and political concerns you rightly point out, but one that responded by direct attack on the automation technology itself.
Then, as now, nuance is hard, and introspective nuance that hits one’s sense of self and sense of purpose is really really hard. I imagine some Luddites saw attacks on equipment as a means of direct action towards political and economic change and equitable distribution of the gains of automation. I also imagine a hell of a lot of them were justifiably scared and unbalanced by a change that shook their economic and social status, and absolutely wanted to hit pause and halt the technological change itself.
I see a lot of breathless corporate bullshit pushing AI in unjustifiable ways, and I see a lot of genuinely transformative power in the technology. I see a lot of very reasonable anti-AI backlash because of that corporate behaviour, and I see a fair amount of unexamined blind anger against the technology itself too.
90% of the time, when someone frames an issue as “taking jobs”, they’re contributing to the anti-technology rather than anti-corporate argument whether they intend to or not. The implication of “taking jobs” is that the jobs should remain, despite a machine being able to do them instead. Sometimes there’s a quality of work argument attached, which is where things get more nuanced again, but often it’s a genuine belief that jobs are a necessity and that blocking technological progress is preferable to eliminating them.
Wealth redistribution and work reform are legitimate anti-corporate arguments around properly allocating the gains of automation for the good of society as a whole, but both require an uncomfortable conversation about (a) the fact they probably won’t happen in a serious way without the kind of societal change that’s historically brought a whole lot of bloodshed as well, (b) they require a widespread severing of the “paid employment -> sense of self worth” link that many people hold dear and that needs to happen before a critical mass of people are willing to seriously push for reform (in a hopefully-nonviolent-but-let’s-be-realistic-about-history way).
You’re going to have to make a stronger argument than just calling people Luddites for not just accepting this carte blanche. Do you actually know anything about the Luddites beliefs and why they were doing what they did?
Why should people be trusting those in positions of power in the tech community?
Imo, calling LLM-tooling “AI” is the best marketing strategy these companies had, and also terrible for how we have conversations about these tools.
When you frame them as “AI” and give them names and put them behind a messaging-like interface, it’s easy to anthropomorphize these tools. But they aren’t doing “intelligence”, they’re doing advanced statistics.
The tools are not “going rogue” and these companies are not “losing control”. Those stories about LLMs “hacking” other companies are a result of having no guardrails, infinite compute, no sandboxing, and unclear goals / instructions. If you prompt an LLM with an objective of retrieving a file from a server and give it unlimited resources and a shitty sandbox, of course one of the methods it might generate could be, essentially, hacking. And if you blindly trust the output of LLMs without supervision, you’ll end up doing hacking.
If these companies or their employees were serious about whatever fears they claim to have, they would be ceasing all operations and doing everything in their power to make everyone else cease operations. Since they aren’t, the only explanations are that they either don’t care that “AI is going to kill us all” OR that they don’t believe it and are saying that as a fear-based marketing strategy. Both, morally reprehensible, imo.
So…no, I am not particularly scared of the impact these technologies are going to have.
I am worried about the commitments by large tech companies to this technology. So far, it seems many companies are taking on lots of debt to support the buildout of the LLM companies despite, from what we know, the LLM companies are deeply unprofitable. Which means prices will need to rise or the tools will need to go away. Additionally, a lot of companies are seemingly very willing to hand over process knowledge to these LLM companies and get rid of or de-skill their own employees. Which again…seems shortsighted if those LLM providers disappear or need to raise prices.
I am also worried about the language we use to describe these tools. We don’t really have the words to NOT anthropomorphize them, and I think that is a huge driving factor in why so many folks get very…irrational about these products.
Edit: Also worried about the way these tools and their users might affect the social contract we all have. Many people seem very okay with saying that material output by these LLM tools is something “they made”. I generally feel the same way about this as I do with plagiarism. Which is not very positive. If someone doesn’t care enough to make something, I don’t care enough to engage with it. And if that person lies to try and trick me, that’s pretty awful.
Personally, I'm retired, so no job worries. I can tinker with AI when I feel like it, and I think it's fascinating.
But like many others, I dislike seeing new AI features in software that I already use. After upgrades, I'm often looking for ways to turn new features off. I'd rather that AI appeared only in new websites and apps, so that people could make their own decisions about when to switch to something that's AI-enabled.
Instead, I have to switch products to get away from it! VS Code is a good example of that; I'm probably going to switch to an editor that's just an editor.
More generally, Artificial Intelligence is a major news story and I think it will somehow affect most people everywhere in the world. Maybe not as much as the pandemic, but it's getting there!
During the pandemic, there were a lot of news stories and political issues that were just about aspects of the pandemic. You could discuss various scientific issues, and how the pandemic affected various countries, and the politics of that, and legal issues, and how the pandemic affects markets, and supply shortages, and its effects on education, and its effects on international relations, and already existing political issues that the pandemic somehow aggravated. It just kinda affected everything, and that made discussing it complicated.
Similarly for AI. I'm worried about polarization where people decide they're either pro- or anti- AI and that predetermines what they think about the various things AI affects. I think all of the following are separate questions, which then divide up into smaller sub-questions, and each of them could have a different answer!
What compensation, if any, should there be for allowing data somehow associated with you to be used to train AI?
Should AI-generated artifacts be considered to violate copyright or other laws? What is the burden of proof? Do you need to actually demonstrate that something is a copy?
What effects do data centers have on local water supplies? Water supplies are local or regional, so this is a local question and the answers will vary.
Similarly, what effects do data centers have on the local power grids? Are the companies that build data centers paying enough for grid improvements? Should they be required to use green energy?
More generally, how much should local governments be compensated for allowing data centers to be built?
What are good and bad ways to use AI to learn new things? How do we encourage students to do more of the good things and less of the bad?
What skills become more or less relevant now that AI can do some tasks?
Will AI have a big effect on robotics, and over what sort of timeframe? What do we expect these robots to be able to do? How will that affect the job market?
Should we expect big effects in healthcare due to AI use?
What changes should we expect from military use of AI?
And so on and so on. Nobody really knows what's going to happen. We are going to need to figure out how to live with high uncertainty for a while.
Along with that uncertainty, I think we should have a wide Overton window. There are many plausible scenarios, some of them quite alarming. I'd like to be able to discuss them without making overconfident claims that one scenario or another is The Future.
There are two sides to this, it has already taken my job to a large extent, but otoh my job has changed drastically. It has gone from implementing features and carefully building a maintainable system to more of coaching an agent harness to produce the desired features. The code has essentially become assembly (nearly no one read or wrote code at that level before LLMs for the last few decades). The skills I've built up over the years are still valuable in nudging the agent in the right direction and having it avoid traps, but overall being a developer/progrmmer is in my case mostly gone.
Yes, there is less enjoyment in developing software this way if you enjoyed the craft. If you are motivated by the output, then now is really your golden era.
My eldest is on his last year of the high school equivalent here. He has a penchant for progamming and he really enjoys it, but I think I will advice him to go into some other proffession and keep it as a hobby.
Via work; claude and via google some gemini (I've disabled all the tracking I could in this one). Gemini is usefull for finding out things and resolving issues and just general advice. I still ask it for references more often than not and it does lie not too infrequently. I asked it what the license was for a project I have on github and it confidently claimed it was MIT. I had not set up a license file for the project yet, and when I challenged it it immediately backed down.
I lean towards net possitive. The economics of working with an agent productivity wise CAN be significant. The issues with energy/water consumption I think can be overcome. There might be some Jevons paradox going on such that we will just increase our use of this tech and the productivity baseline for regular office jobs will be higher... then again I kind of find most office work to be "bullshit jobs", sooo being more efficient bullshit jobs has little real world benefits. In the end we all basically need food, shelter and a sense of being valued. AI as is don't really contribute to any of those in a meaningful way (maybe negatively in the "sense of being valued" part).
Not at all. We haven't changed that much in the last 200 000 years. We've just added more rituals :)
Yes.
Does it bother me? No. I'm quite secure in my own capabilities (and lack thereof). Other people will be smarter than me, some will be dumber. If these other people are instead AI, it's not that much different. The difference is in speed, not in kind.
Being a software developer this has already impacted my sense of professional worth, but I am not my work nor my role at work. That is not where I derive my sense of worth as a person. Hobby wise I think I'll just keep playing pen and paper RPGs with my friends aka. the cheapest hobby known to man (maybe excluding singing).
I don't think this is realistic to be honest... Humans will continue to human with or without machines... I'm not too worried about alignment issues.
Neither. What will be will be, it will likely involve a bit of suffering and adjustment so short term it might be a bit rough. Long term I'm not too worried. On geological time scales we're all dust.
I think the problem is alignment, but not for machines, rather, for us. Much of the conversation around the dangers of AI has been around the threat it poses due to its capability. Indeed, the analogy is frequently made to nuclear weapons, of which there has, since their development, also been a great deal of anxiety for much the same reason.
This seems misplaced to me. The danger then and now is not from the machine, but from us. Specifically, the danger is from the philosophy of materialism and the death of religion as a motivating force for human behavior. Not unlike nationalism, racism, communism, and other philosophies that have emerged in the wake of the loss of religion in society, the current strain of materialism, call it "computerism", creates dangers for everyone involved using the tools available at the time because these notions are absent of the qualities that religion imbues in a society.
While these vary in emphasis from one adherent to another, every religion has taught the importance of hope, faith, charity, submission, kindness, forgiveness, forbearance, fairness, and so on. Having abandoned the sources of these values at the turn of the 20th century, where have we found them in the materialist ideologies that came in their wake?
I bring all this up because I think the danger we perceive in AI originates from our own conception of human nature, a conception that has become divorced from every praiseworthy characteristic to which one was to aspire in a former age. The problem is not that the machines are mis-aligned, it's that we are. If, as some have predicted, the machines were somehow to precipitate the extinction of humanity, it would be because we had come to see others only as selfish antagonists with no meaning or soul or existence beyond their time in this life. It would be because lost the faith in religion and each other, not because of some nefarious LLM.
The solution is not, then, banning AI or something like that, but it is us. It is our ability to build bonds of fellowship and faith, whatever that might be, with those around us and talking about how there is more meaning to life than ones narrow material concerns. That refining our character and striving towards spiritual ends is not just important, but is the most important aspect of our lives.
I think the moral aspect of our lives, not the computational one, is what will define how much harm or benefit we obtain, as a society, from technological advancement.
Perhaps I'm misunderstanding something here, but are you saying racism was invented in the 1900s because of a lack of religion? This all sounds extremely ahistorical and ungrounded.
This is glib, but why not both? I expect there to be both malicious use of AI and new kinds of accidents when people didn't take sufficient precautions. And like the HuggingFace incident, I expect some of these accidents to be weird.
Cannot help but reply to this quote from you because it's hard to engage with your argument re current morals without doing so. Are you serious about this? What religious history have you read??? Because sure, many religions support charity, hope and kindness etc, but the problem is in the execution and in the evaluation of who is worthy of said kindness, and who is a filthy heathen that either needs to convert and slot in at the bottom of the believer pyramid, or who needs to purged with fire and blood.
I don't think the decline of religious influence has anything to do with modern morality. If you believe that then I seriously wonder what explanation you have for the many many MANY atrocities and wrongs perpetuated by religious people historically.
I recognize the potential of AI to change a lot of things, but I wouldnt really say Im worried. I just think of it as: the world is always changing, society is always changing. It was never really my choice for which way things go, I just need to adapt to whatever ends up happening.
Mechanization really changed society a lot. Computing did as well. But in both cases, new niches formed for people to fit into even as others disappeared. I dont think humans will ever be obsolete because our importance to eachother extends beyond things like skills or knowledge or even emotional connection or comfort.
Im not at all worried about my job being replaced by AI, because you cant replace being unemployed. But even if the career I am trying to pursue is swallowed up by automation, thats fine. I was in the middle of a career pivot anyway. Ill just look for something else to commit myself to.
I am personally incredibly excited. We had a bit of unlimited Astra at work (programming) and it was the first time I really enjoyed using AI to code. It seems very accurate, pretty smart on its own. It elevated my code reviews to a next level of competency, spotting some very tough nuances and understanding the purpose and design behind the feature from just reading the pull requests, with no execution or tests.
I also tried one of the modern AI arenas and got genuinely very impressive results with some quite fun browser games being created from just one prompt. It also demonstrated the huge difference between latest models and even just one generation older.
I think I'm finally going to get a personal subscription and start building stuff. After years of being stuck in a creative glut this was like a breath of fresh air.
I'm so glad that all of the doomer predictions about it being just slop, about them running out of data to train on, about the models hitting their peak were so wrong. The progress is evident and I am hopeful. I think it's also great that there's such an arms race of competition which is what drives these companies to spend enormous cash in order to advance this field. And I'm here to reap the rewards.
Will it take my job? Very possibly in a decade's time. That's why I'm now switching my focus into mastering these tools and the process of creation.