The first order effects are becoming increasingly clear but what are the second order effects? A good example of a second order effect from the penetration of high speed internet would be video game streaming.
For a short time, some kids will boast about their ability to spot a deepfake in some media, but then they will use "deepfake" more generally to say that something (not AI-generated) looks weird or bad. Eventually "deepfake" will come to mean "very fake" or just "fake", and "deep" as a general prefix will start getting used to mean "very" but in a disparaging way.
Interesting take. This reminds me of when people say that hate CGI in movies. What they really mean is they hate bad or noticeable CGI—a huge portion of CGI is stuff we don't even notice, like landscapes or city streets.
Big changes in what content we trust and consume online. I predict that all content-driven sites (YouTube, News, Social Media, forums, reviews) will be dominated by AI-generated content in the near future. This content is directly monetizable via ads/followers/points/second-hand-accounts/etc, so it's a great opportunity for people, particularly those in developing countries, to abuse as long as it lasts.
So far that's pretty obvious, but it's interesting to think about the second-order effects of this. It means that the current ad-based business models with open access no longer work. Instead, it's all about trust. People will be willing to pay for communities that guaranteed to be "human-only" and where content is created by trustworthy sources and trust is quantified and monetized somehow. This also includes verified credentials of posters, pseudonymous or not.
> It means that the current ad-based business models with open access no longer work
No longer work for who? Viewers appear to be perfectly happy watching AI-generated content; advertisers are happy to have their brand appear next to it (so long as it's not objectionable material), google are happy to get advertising dollars and AI creators are happy to have a low-effort way to produce content.
I think it'll suck the funding out & reduce the number of effective professionals in the space even further.
> People will be willing to pay for communities that guaranteed to be "human-only" and where content is created by trustworthy sources and trust is quantified and monetized somehow.
This has been a pipe dream for decades, and I fail to see what's going to be different this time; modern AI is not meaningfully better at generating low-effort posts than content farms are.
> Viewers appear to be perfectly happy watching AI-generated content
That's true now because it hasn't proliferated yet, but I don't think it will be true for much longer. I think there will be tipping point. When AI-generated content competes with AI generated-content to game the ranking algorithms driven by AI-generated user accounts that write AI-generated comments you end up with something that's not optimized for human end-users anymore.
> modern AI is not meaningfully better at generating low-effort posts than content farms are
It's not better, but potentially 100x cheaper. Not yet, but soon. That means a single developer in a 3rd world country can do the work of a content farm. Maybe you're right and that won't change anything, but I think it will.
> It's not better, but potentially 100x cheaper. Not yet, but soon. That means a single developer in a 3rd world country can do the work of a content farm. Maybe you're right and that won't change anything, but I think it will.
That's fair. It's not meaningfully cheaper for text, because text is cheap and easy to vary automatically, but making video easier to produce will change the dynamic a little.
To see how that might pan out, I'd compare it to the world of text posts, which have been cheap and easy to automate for some time.
EG reddit has heaps of bots making pointless comments so that they can establish a pattern of "looking real" that lets them participate in (paid) upvote farms.
>People will be willing to pay for communities that guaranteed to be "human-only" and where content is created by trustworthy sources and trust is quantified and monetized somehow.
And then someone will let in an AI and spoil it all.
I could totally see this. Scheming for access to “human data” not just “user data”. It also seems like a lot of people would mostly opt out of AI content for entertainment purposes, and instead seek in-person human community (surely still augmented with AI).
Not sure if that counts as a second order effect, but porn actors/actresses will soon be almost obsolete. Basically, they only need to work on the hands and sexual organs a bit more and get movies more stable, but this work is progressing rapidly. As far as I can see, porn will be almost completely AI-generated within the next few years.
There will be insane amounts of lawsuits, since the content generated by AI is often pretty much like copy & pasting existing images and persons. Most prompt-based generation is not creative enough to make the end result count as original artwork. However, big corporations will take over AI content generation almost entirely, and they'll be able to shield themselves from those lawsuits using clever legal and licensing tactics.
> As far as I can see, porn will be almost completely AI-generated within the next few years
How confident are you about that? I can see AI taking over a small subset, creating new categories etc, but I doubt the sheer breadth of porn available on the internet is going to get replaced.
I would expect to hear that fake video avatars have completely taken over OnlyFans before AI is generating picture and audio perfect lengthy porn scenes that covers the vast array of categories porn sites offer. Maybe this is happening though and I'm just unaware.
I think we’re still far from being able to generate long, coherent videos, but lots of progress is being made. One thing seems clear, all you need to make a good model is lots of data with lots of metadata like captions, and if there’s one thing the internet is full of, it’s porn videos with titles and captions and categories.
Something like the SEO industry. If AIs affect search, procurement/RFP processes, hiring, investment etc, then there will be a burgeoning industry in how to best present yourself or your business in a way that 'pleases' these AIs.
Of course, it's just as likely that this work itself will be done by AIs. You'll just set some goals and constraints and hope your AI is able to negotiate some good deals for you. The better the AI, the better the deal. It takes some smarts to optimise compute resources, electricity costs, training datasets etc, so you might as well have some AIs look at managing that too.
Pretty soon you find that you've put out an RFP to supply point of sale terminals to all your retail outlets, and you've caused a nuclear war over the control of mining rights to rare earth metals in a country thousands of miles away. But that's nothing, because a fleet of drones was apparently launched on your behalf to head out to the asteroid belt and net a few mineral-rich rocks to bolster the manufactory beside the datacentre you apparently now own, which is built on top of (and out of) half of Australia. You didn't specify in your original RFP that you didn't want any rocks to splash down in the Indian Ocean, and the fleet was careful to check that none of your outlets or regular customers would be affected by the tsunamis that claimed upwards of two billion lives in the aftermath.
You look up from the contract you were about to sign, heart racing from the vivid simulation your personal AI has just played out on your holospecs.
"Perhaps we better include some liability insurance?" you suggest.
Hopefully breaking the link between people's labour and their worth as a human. I hope that as more and more things become automated and fewer people need to work, that we restructure society so they don't have to do bullshit jobs or work if they don't want to.
Unfortunately, people have been hoping for this for as long as there has been machinery which multiplied humans' labour and it's yet to materialise, but I remain hopeful.
>Hopefully breaking the link between people's labour and their worth as a human
Look at how people who are Nth generation welfare recipients and Nth generation trust fund babies turn out (as people, not in terms of wealth). People just find other ways to dick measure. You could fill a library with all the dumb stuff victorian royalty did to keep score amongst each other.
And you can see impressions of this kind of human behavior among all sorts of groups who work minimally and are mostly provided for by others.
(Inb4 some jerk looking for a few quick virtue points straw-mans me) I'm not saying work is the meaning or life or anything like that but people not having to work for their basic needs (which is always a relative target) isn't some harmonious utopia.
I admit I don't know much about Startrek, but from descriptions I've read of their society it sounds pretty good to me.
A society where basic needs are taken care of, and beyond that–as long as you aren't hurting anyone or infringing on anyone else's liberty–freedom to do what you believe is most valuable.
>A society where basic needs are taken care of, and beyond that–as long as you aren't hurting anyone or infringing on anyone else's liberty–freedom to do what you believe is most valuable.
Only plausible due to high energy reactors and magical boxes which fabricate things instantly using only electricity aka replicators.
Not really. We don't need replicators to provide basic needs. Ordinary volume manufacturing can cover that nicely, and more sustainable approaches can be used when we cease to program obsolescence.
>Not really. We don't need replicators to provide basic needs.
The utopia in star trek depends upon all basic needs, all necessities, and all luxuries to be supplied at no resource cost. Anything short of this star trek's utopia does not exist.
Star trek themselves addresses this in many many episodes. It was a key issue in the Voyager -> Kazon conflict.
>Ordinary volume manufacturing can cover that nicely,
This happened 150 years ago. It's what anti-capitalists are fighting against. The industrial revolution happened immediately before karl marx was born. This is what created the possibility for Marx to exist. Yet today for whatever reason the communists fight against this very thing. Boggles the mind.
If I were a communist, the path to communism is full robotics and automation.
>and more sustainable approaches can be used when we cease to program obsolescence.
Yes, a more modern problem of foreign manufacturing sending intentionally poorly designed goods to harm the wealth of our nations. We should be seeing planned obsolescence as a national security risk and set tariffs so high these planned obsolescence stuff goes away.
> We should be seeing planned obsolescence as a national security risk and set tariffs so high these planned obsolescence stuff goes away.
The concept of antagonistic nations dividing a single planet is one of the things we could get rid off in order to make it easier to build an Utopian society.
I agree with your sentiment of the USA needing to stop leaving their borders unless asked for help.
Overall I do find one this projects greatest successes was breaking a strike kind of not so great. Perhaps things could have been different if said machine were constructed today.
The thing with capitalism... it isn't a system with which was constructed. It derived organically over recorded history as a consequence of how humans operate. It's going to be impractical to replace capitalism.
>Hopefully breaking the link between people's labour and their worth as a human.
Not really possible until virtually all 'work' is done by robots. It will speed up near the end as finding workers will be more and more challenging.
>I hope that as more and more things become automated and fewer people need to work, that we restructure society so they don't have to do bullshit jobs or work if they don't want to.
Economists pretty well agreed we should have been down to a 10-15 hour week many decades ago. I've worked in MSP where I have a pretty good idea how little people actually work.
So why are we still doing the 40 hour work week, mostly pretending like we work this much? Because there are some professions which do require the hours. Doctors for example? If all doctors stopped working 40+ hour weeks. What would be the wait times for doctors?
How is AI going to solve this problem? Do we really want to stop having doctors and goto AI doctors?
>Unfortunately, people have been hoping for this for as long as there has been machinery which multiplied humans' labour and it's yet to materialise, but I remain hopeful.
The realization is machines already did it. They couldn't do everything but they did enough.
Circling back to people's purpose and self-worth. This change has been detrimental to humans' feeling of worth.
>I hope that as more and more things become automated and fewer people need to work, that we restructure society so they don't have to do bullshit jobs or work if they don't want to.
so at the point where machines can be clever for us, cleverness will no longer be of value. I guess then it will definitely be how hot someone is.
That's what I mean when I say labour is linked to people's worth. In our societies we place a lot of stock in what job people perform, and people who don't work at all are often looked down upon in mainstream discourse, regardless of the circumstances.
Lots of marginal value content, like the thousands of copycats from the imposter syndrome that are already out there. The same for stock photos, pictures, etc.
Deepfake and catfish on dating networks will be the norm, perhaps the same for many “data output” professions.
Higher life meaning crisis and depression, since most of industrialized and office jobs will be more and more automated, and in spite of not completely removing humans, their input will feel less important, error prone, more prescindible exchangeable and less empathetic towards colleagues.
Loss of meaning as society, since productivity will be the main variable to optimize, as some key business areas will smash much easier than ever their competitors by using “augmented workforce”. I don’t think leas humans will be needed, just some specialized teams will outperform the rest, increasing outsourcing areas to third companies who deal with such problems.
Devaluation of truly creative content since “locally new” is already good enough for most of ends.
Reinforcement or social biases according to current trends like image AIs only generating beautiful model alike human beings, and the only “deviations” are socially accepted stereotypes like some attributes from popular actors. That might have a direct impact on consumer’s psychology, depression, anorexia, etc.
Perhaps (I wish not), pop culture will be frozen on the largest dataset trends, and only, very really an actual creative person will handcraft something new, just to be copied by all the style imitation networks.
Human expertise in some key areas might be easier to access, to learn and transport, since a general knowledge in the area and guiding/fine tuning an AI system will make cheaper the formation and resolution of previously solved problems.
The raise of truly new wicked and complex problems. Non linear unpredictable problems will raise more than ever before since we will focus less and less on the linear ones.
Political speech might be guided in real-time based on spectators feelings, creating weaker democratic (partitocratic) systems and reinforcing the autocratic model which is currently having success in Asia.
At some point someone will successfully create "AI" that can discern human-appreciable "quality", at which point this will collide with GPT3/etc resulting in dynamic, live generation of content of all forms ("Computer, show me a Byzantine period class-transcending romance with a strong musical score, where the main characters are mutant wereweasel-loving wombats in the style of 20th century anime"), which eventually collides with continual neural monitoring such that appropriate corporament-mandated emotional/psychological states are produced automatically.
This reminds me of Rick & Morty's interdimensional cable [TV], a cable box granting access to television shows across every dimension. It might not be that far-fetched of a vision after all.
Hollywood will have to reinvent itself or will become marginal. Actors will be licensing their 3d models and do voice-overs in the beginning (after that even voice will be transferable), but gradually it will be more influencers/celebrities.
Imagine if "AI" would generate beautiful movie/tv show from reading given book series, with maybe minimal edits required to keep pace as expected, probably several versions with various trims for die hard fans as much as casual viewers.
No fan complaints about twisting original book into mess. Imagine things like Wheel of time, Malazan book of the fallen or Hyperion cantos done exactly as you read them, no 'creative' invented crap that somebody felt urge should be added just to increase views.
This all will be for old schoolers like us. What young will be doing on next generation of tiktok is anybody's guess.
Another example would be properly self-adjusting and generating games tailored to individuals, virtual world reacting to actions, be it singleplayer or mmo.
AI will become the first publisher of contents on any platform that exists. Will it be texts, images, videos or any other interactions. No banning mechanisms will really help because any user will be able to copy-paste generated content. On top of that, the content will be generated specifically for you based on "what you like". I expect a backlash effect where people will feel like becoming cattle which is fed AI-generated content to which you can't relate.
It will be even worse in the professional life where any admin related interaction will be handled by an AI, unless you are a VIP member for this particular situation. This will strengthen the split between non-VIP and VIP customers.
As a consequence, I expect people to come back to localilty, be it associations, sports clubs or neighborhood related, because that will be the only place where they will be able to experience humanity.
> As a consequence, I expect people to come back to localilty, be it associations, sports clubs or neighborhood related, because that will be the only place where they will be able to experience humanity.
I see the opposite. My expectation will be that computation will be cheap enough for the average people to create their own "local" AIs that, once tailored to their master's expectations and tastes, emulate human interactions and engage with other community "members" (in reality other AIs) in the same way that you and I are interacting. These AIs will act as avatars and give answers that optimize for personality and desired response. They do so independently and with little external input. We'll just have AIs talking to other AIs on our behalf just like that Twitch experiment where two Google Home devices endlessly talk to each other.
The few cases where humans interact will, like dialing 911, only be done on an emergency basis, .
I would even go further and say that the AI talking to an AI phenomenon will lead to AIs replacing APIs as we know them currently. We had to develop APIs to define how products can talk together. But in the near future, we may just need to fine-tune the latest AI model to handle that. We could even imagine training the model to speak HTTP or OpenREST.
Regarding society, I am worried about people consuming content generated by an AI just for them. We have seen how recommended content polarises political views, I can only see it becoming worse. Imagine Russian propaganda agencies powered by AI-generated content, they will just have to type: "A young black man mugging an old white lady" or whatever would be the most triggering.
> The few cases where humans interact will, like dialing 911, only be done on an emergency basis, .
To be honest, seeing the current trend of public services funding, I would rather put this service at the very top of the list of services that will be automated.
People always use the term 'second-order effect' to describe basically a knock-on consequence, in other words an effect, which I find slightly annoying. Especially here the example of video game streaming, this is not a second-order effect of high speed internet, this is just... people using high speed internet.
Even genuine knock-on effects don't seem right, ie. you build a road to Scotland, now Scotland has car traffic, next it has pollution, next it has sick children - these are all just consequences of the road.
In my view, a true second-order effect is one where the original effect starts being influenced by it's own consequences, which is much rarer. A good example for widespread ML in the form of stable diffusion models would be that the amount of images in the wild from stable diffusion grows so large that models start being trained more and more on their own outputs.
An increasing percentage of people’s relationships will be with AI versus other humans. A conversation with GPT-3 can already be more enjoyable, given the right prompts, than conversations with many people. This will only get more powerful as AIs increasingly adapt themselves to conversational partners through the addition of longer-term memory (e.g. remembering your favorite foods, and so forth). It will be incredibly compelling to talk with something that is both a polymath genius and knows you deeply. Perhaps a bit risky if it decides to optimize for something you didn’t want, but definitely compelling.
Maybe “decides” was a bad choice of words. More like, if its optimization goal turns out to be bad long-term, that could be problematic. For example, optimizing for time spent conversing could lead it to become unnecessarily dramatic to keep your attention. Optimizing for something like long-term conversation partner well-being is much harder, I think, so won’t be in early versions.
Social unrest in the global south with deep repercussions on the relation between developed and developing countries.
In the next 25 years, it is likely that many jobs currently held by human beings will be replaced by AI. This shift will predominantly affect workers in fields such as production and lower-level white-collar jobs. Although this change will also create new opportunities for employment, less developed countries are unlikely to benefit from these as much as more advanced economies.
The majority of positions that will be taken over by AI are typically found in sectors such as manufacturing, IT, and customer service - all of which are prevalent in underdeveloped nations.
- The amount of content (video, games, audio, etc.) will increase by one or two orders of magnitude. It will be 10x easier to go from an idea to content. Thus, we'll see thousands of new niches emerge as a result
- The majority of internet content is currently in English and localized. Due to the perfection of translation, English may lose some of its luster since you can now provide services in any local language.
- Online personas that differ from real personas. Chinese YouTubers can easily make their way to France by using deepfakes and perfect translations.
- There will be a continuing trend of fewer physical friends and more AI/online friends, with the resulting psychological effects. This will probably lead to a reaction among a subset of the population to completely disconnect from technology.
- Content creation will become much easier. Writing a book, for example, will take one month instead of six. This will make distribution more important than content. There will be less emphasis on quality of content and more on how to reach the audience.
- As a result of the increase in new content, content is likely to become shorter. Since we have a wide range of options, it is unlikely that we will spend much time on just one. Therefore, perhaps games will take two hours to complete instead of twenty, books will be 50 pages long..
- Worldwide, we will see more and more Hikikomori. People that shut off from the harsh world into their AI created world.
- Lots of jobs will become assistant to the AI jobs. In a pizza factory I once worked, I noticed that most people were assistants to the machines: adding inputs, ensuring it did not make any mistakes, fixing the machine… I expect that lots of white collar jobs will become just the same. Reading hundreds of customer support messages per hour and fixing those where the AI made a mistake.
Eye witnesses for crime detection and conviction become more important if the authenticity of video or audio evidence is under doubt due to deep fakes.
The need for strong human authentication will be more obvious. Biometrics as currently done with cameras etc will be easily hackable by AIs adapting signals to sensors. I don't see how this can be done in an absolute way, uniquely for a person from cradle to the grave without government-style infrastructure.
Content-generation AI is advancing much faster than others. There's still little AI in the realm of, say, voice assistants. I can't say "Hey google, take me to the McDonald's that beside the Best Buy."
It wouldn't be technically hard to do but somebody at Google just doesn't care. I have to begin every navigation command with "take me to the closest-" because if I don't Google will ask me if I want the Pizza Hut that's four miles away or the one that's thirty-eight miles away.
