When this current bubble pops, and it will, the tech can be developed to its full potential after that. Right now, the market is 99% snakeoil.
It depends on what bubble one’s referring to. The tech itself isn’t going to ‘pop’ - in many cases the capabilities will probably outpace the current promises given the compounding rates of improvement. This isn’t like past tech buzz cycles which is part of why there’s a lot of questionable predictions regarding it.
Yeah, snake oil bottom feeders will gravitate to any buzz they can attach themselves to. But the barnacles don’t steer the ship. The market of snake oil will dry up as it always does, but that’s largely because their primary industry is selling snake oil, not whatever they change the label to.
The reality remains that Meta, Google and Microsoft will not profit if the world’s problems are solved.
Not really. Microsoft stands to make a killing just running these models on Azure as their sole line of business if AI ends up as successful as it may prove to be. Google divesting more from ads might prove to make them less evil. Meta would be evil in this space if not for the fact that because they started off late they’re the biggest driver of open AI development right now and arguably the biggest funder of any hope of counter-corporate AI existing.
It’s easy to regard companies as monoliths. And while it’s generally true that a corporation, especially large public companies, will end up trying to optimize around short term gains even at the cost of long term consequences or social evils, it isn’t necessarily true that public good is always at odds with capitalist self optimization in all things. So it would still be a win for Microsoft if AI allowed for the public good as long as they could ensure that AI was running on their servers and they could attempt to maximize their margins as much as the market allowed for before net gains decrease. And any corporation smart enough to focus on longer term gains is going to be one that’s going to actively try to avoid excessive public harm as your longer term revenues aren’t going to go up if your customers die or go homeless, etc.
Also, the researchers themselves have certain aims and if their parent company doesn’t align with those aims, they may take themselves and their significant value away from that company. For example, Meta was only suddenly “open AI friendly” and then a major player after literally half their AI team quit for greener pastures.
Unfortunately, the risks are just as massive. Everyone just seems to be blinded by the “new shiny” and refuses to see any negatives…
While weighing risks, it’s also important to weigh opportunity costs.
Also, I’m not sure who you are interfacing with, but in my experience it definitely seems like the majority of people are fairly bearish regarding AI (there’s a number of reasons why I think that’s the case, but it’s still a significant majority). These days positivity regarding AI that isn’t in the context of a snake oil sales channel is a rarity in most public discussions.
Also, I’m not sure who you are interfacing with, but in my experience it definitely seems like the majority of people are fairly bearish regarding AI
(I am going to run on the assumption that you meant bullish instead of bearish? I’ll clarify my view, just in case.)
I have been in IT Security for a very long time. People in my industry are usually pulled in to clean up the mess after the party is over. It’s not entirely bad as it pays my bills and in general, I am bullish on AI in the long term and extremely bearish in the short term.
Thinking of a company as a monolith is easy. The sum of its parts must always equal profit. Sure, a company can have good parts and bad parts but one side cannot justify the other. (That could be a much deeper philosophical discussion for another time though.)
What pisses me off now are the number of people and companies that not weighing the risks or have a full understanding of what they are doing in the rush to implement. To use an old phrase: “Just because you can implement something doesn’t mean that you should.”
From my perspective, many companies are just shoving all of their sensitive data and our PII into a magic box to see how many dollars it poops out. It’s sidestepping all of our data protection laws because it’s going to be years before the laws are adjusted. (I simplified the complexity of ANNs and how data is stored in them, but my point remains.)
Even worse, companies are using the magic of AI to slurp even more of our data either willingly or by forced updates. A seemingly benign example of this is Microsoft integrating AI into notepad. Think about it: It is the most common temporary space on a PC where the average user stores highly sensitive and confidential data. Google is going to integrate Bard into messages so it can ingest your entire text history, personal or not. (Point to point encryption and other controls become irrelevant then.)
This particular brand of snake oil is exceptionally potent because products like ChatGPT introduce such high levels of illusion.
Edit: With all of that said, I really do respect your points and where our opinions differ. I also realized I am allocating too much time to this discussion and have a headache now. I’ll gracefully back out now, but that kinda sucks. This would be a much more fruitful discussion in person where we aren’t limited by typing.
It depends on what bubble one’s referring to. The tech itself isn’t going to ‘pop’ - in many cases the capabilities will probably outpace the current promises given the compounding rates of improvement. This isn’t like past tech buzz cycles which is part of why there’s a lot of questionable predictions regarding it.
Yeah, snake oil bottom feeders will gravitate to any buzz they can attach themselves to. But the barnacles don’t steer the ship. The market of snake oil will dry up as it always does, but that’s largely because their primary industry is selling snake oil, not whatever they change the label to.
Not really. Microsoft stands to make a killing just running these models on Azure as their sole line of business if AI ends up as successful as it may prove to be. Google divesting more from ads might prove to make them less evil. Meta would be evil in this space if not for the fact that because they started off late they’re the biggest driver of open AI development right now and arguably the biggest funder of any hope of counter-corporate AI existing.
It’s easy to regard companies as monoliths. And while it’s generally true that a corporation, especially large public companies, will end up trying to optimize around short term gains even at the cost of long term consequences or social evils, it isn’t necessarily true that public good is always at odds with capitalist self optimization in all things. So it would still be a win for Microsoft if AI allowed for the public good as long as they could ensure that AI was running on their servers and they could attempt to maximize their margins as much as the market allowed for before net gains decrease. And any corporation smart enough to focus on longer term gains is going to be one that’s going to actively try to avoid excessive public harm as your longer term revenues aren’t going to go up if your customers die or go homeless, etc.
Also, the researchers themselves have certain aims and if their parent company doesn’t align with those aims, they may take themselves and their significant value away from that company. For example, Meta was only suddenly “open AI friendly” and then a major player after literally half their AI team quit for greener pastures.
While weighing risks, it’s also important to weigh opportunity costs.
Also, I’m not sure who you are interfacing with, but in my experience it definitely seems like the majority of people are fairly bearish regarding AI (there’s a number of reasons why I think that’s the case, but it’s still a significant majority). These days positivity regarding AI that isn’t in the context of a snake oil sales channel is a rarity in most public discussions.
(I am going to run on the assumption that you meant bullish instead of bearish? I’ll clarify my view, just in case.)
I have been in IT Security for a very long time. People in my industry are usually pulled in to clean up the mess after the party is over. It’s not entirely bad as it pays my bills and in general, I am bullish on AI in the long term and extremely bearish in the short term.
Thinking of a company as a monolith is easy. The sum of its parts must always equal profit. Sure, a company can have good parts and bad parts but one side cannot justify the other. (That could be a much deeper philosophical discussion for another time though.)
What pisses me off now are the number of people and companies that not weighing the risks or have a full understanding of what they are doing in the rush to implement. To use an old phrase: “Just because you can implement something doesn’t mean that you should.”
From my perspective, many companies are just shoving all of their sensitive data and our PII into a magic box to see how many dollars it poops out. It’s sidestepping all of our data protection laws because it’s going to be years before the laws are adjusted. (I simplified the complexity of ANNs and how data is stored in them, but my point remains.)
Even worse, companies are using the magic of AI to slurp even more of our data either willingly or by forced updates. A seemingly benign example of this is Microsoft integrating AI into notepad. Think about it: It is the most common temporary space on a PC where the average user stores highly sensitive and confidential data. Google is going to integrate Bard into messages so it can ingest your entire text history, personal or not. (Point to point encryption and other controls become irrelevant then.)
This particular brand of snake oil is exceptionally potent because products like ChatGPT introduce such high levels of illusion.
Edit: With all of that said, I really do respect your points and where our opinions differ. I also realized I am allocating too much time to this discussion and have a headache now. I’ll gracefully back out now, but that kinda sucks. This would be a much more fruitful discussion in person where we aren’t limited by typing.