Will wealthier people access better AIs?
Today, billionaires and middle-class everyday people use the same phones. Will that be the case with AIs?
Assuming everyone gets access to the same quality of AI, how will rich people maintain their compounding advantages over the non-wealthy?
The first way, is by having more computing resources. The same AI, but with more computing resources, is going to come up with equal or higher quality results.
The second, is being able to requisition more physical resources. The same AI can help a wealthy person set up a business, by quickly acquiring physical and IP assets, or labor, vs. someone without the wealth to do that.
The third, is by having better information. Small differences in information, such as having a market's trading history just a fraction of a second before someone else, can translate into a lot of economic power. Wealthy people can pay for better information, or put systems into place to get better information.
The fourth, is risk tolerance. Even with an AI helping someone making choices, some choices with the highest expected return also come with the highest volatility or risk. Someone with resources can tolerate a lot of individual risks. But a low resource person will have to play it safe and forgo those opportunities.
Conclusion: The efficiencies delivered by AI will intensify existing compounding effects, and the inequality those already generate.
Even if AI access was somehow kept even.
source: https://nymag.com/intelligencer/2023/03/on-with-kara-swisher...
Swisher: One of the excuses that tech always uses is you don’t understand it, we need to keep it in the back room. It’s often about competition.
Altman: Well, for us it’s the opposite. I mean, what we’ve said all along — and this is different than what most other AGI efforts have thought — is everybody needs to know about this. AGI should not be built in a secret lab with only the people who are privileged and smart enough to understand it. Part of the reason that we deploy this is, I think, we need the input of the world, and the world needs familiarity with what is in the process of happening, the ability to weigh in, to shape this together. We want that. We need that input, and people deserve it. So I think we’re not the secretive company. We’re quite the opposite. We put the most advanced AI in the world in an API that anybody can use. I don’t think that if we hadn’t started doing that a few years ago, Google or anybody else would be doing it now. They would just be using it secretly to make Google search better.
Technology does not, in itself, create all the advantages. I likely have the same model iPhone as Bezos but his contact list probably includes much more powerful people. Similarly, we could have access to the same AI but he’d be positioned to derive much more leverage from it
I’m sure a super premium category will emerge, but the 80-20 rule likely applies here.
I suspect these models will usually have a "free", highly limited, option. It's the perfect bait! (worked for me)
A few hours a year of compute seems a realistic median usage, so you can do the math. But it's not much. Certainly a premium AI that takes longer to process will cost commensurately more. But again, likely 80-20 rule applies. These more advanced AI aren't likely to be orders of magnitudes better than the cheap/common ones.
It will likely also be common to tier usage such that you only employ more expensive/computationally intensive models when needed. GPT-3.5 can already adequately cover a large number of use cases.
What path do you see that could make open models competitive?
I think a better analogy (or perhaps a more specific one, since you did mention "web services") would be computing services, i.e. rich startups with hundreds of thousands in credits and funding vs. single bootstrapped founder with a little bit of cash.
Indeed, and that's kind of my point. Even in the most egalitarian product in history, a bit of a wealth gap exists. It's inescapable. That $60 game may not cost $6,000, but it won't run on cheap hardware. And it'll probably depend on a good, high speed Internet connection to acquire in the first place...
Current AI is either relatively low cost but centralized, or very expensive to run locally. I believe AI will be more egalitarian than software in general (as Stanford Alpaca showed), if a lot of work is done to make inference at the edge practical.
The open community shouldn't lose sight of this.
It's probably worth anticipating and finding a way to mitigate it. In the early days of computers and the internet, schools and libraries had computer labs that were a decent equalizer for kids who didn't have them at home. Maybe we should be thinking about how to make early AI similarly accessible.
Trading your private info for societal gain is a proven model, it'll now extend to your..voice patterns?, and so on.
Software doesn't really lend itself to custom elitism because the marginal costs are usually close to zero.
Don't assume the rich have just one phone. It is common to compartment with several different phones for different purposes. One you only give to family/friends, another for strictly business topics, another for social media, another for giving out publicly on business cards, and the list goes on.
Half a dozen phones's obviously a bit much, but if a phone gets dropped and breaks, the IT person should already have a replacement ready to go, no waiting around being phoneless while someone goes to go to the Apple store.
Do they? I assume they would have at least additional security features and security monitoring given they're a higher profile target than the average person.
running your own will allow you to write anything