Clojurians Log v2

The world depends on resources that are finite, and thus, we are in a serious energy and environmental crisis. Gen AI is the straw that broke the camel's back. I think technology can be used to improve energy efficiency (homes/factories), circular economy, sharing economy, make our cities/vehicles smarter and cleaner and so on. I also believe that AI/ML models can play an important role in this “tech for good” sector.

"social and environmental" seems to be a pretty big bucket. I think the biggest social problem we have is coming to a social agreement about how these agents will and won't behave around our children. Seems that'll be the new argument for decades to come. Wrt energy/pollution, yeah I think the progress of technology is the best chance we have of improving inefficiencies and cleaning up dirtiness. I think forcing people into performative rituals that don't actually fix the problems will actually push people away from solving the problem. What I worry most about is us building an economy on these memento casinos where individuals and companies are beholden to AI slot machines where the house always wins. It's hotel California future we don't want to go down, I think. We need to wrest the thinking kernel out of the data center, bring it local, and use these data providers as an outer cache with the many trillions of tokens. A future where AI does not go local is truly dystopian IMO

In general though, I think this is a safe space for you to talk about greener practices. Did you have anything in mind? There's a lot of emotions around the topic and if hyperbole rises the mods might shut it down, but in general the topic is permitted here, from what I've seen

Wrt kids, it's interesting, we're going to want them to lie to kids, under a certain age. Talking about whether Santa exists might be off limits! Lol And like, if a toddler walks up to a robot and starts bashing it in the face... The robot should not say, "haha, it's okay, it doesn't actually hurt me, hit me as hard as you want, haha!" It needs to say, "ouchie, ouchie, haha, that would hurt, haha, ouchy ouchie."

And then you have all these layers of honesty and make believe that children develop through. And these agents need to know how to behave at each layer of the development, in a way that is cross culturally acceptable

But obviously, if those protocols become the subject of public debate, it forces into language and law what the definition of a human and its purposes are

We need to avoid creating a closed system of possible future purposes, incapable of evolution, while not cutting off the branch we're hanging on

It's like, everybody is a libertarian except when it comes to the children. Until they're free to diverge from traditional human purposes as an adult, we disallow other adults from interfering in their traditional development, until such time the child is mature enough to choose between the more complicated influences. The most extreme libertarianism would place adult level sovereignty at the point of conception, as if freedom to opt out of tradition is always available to the child, which doesn't work developmentally. A human is definitely a multi stage metamorphosing animal that depends on fictional worlds throughout its development. The freedom machinery has to be bootstrapped

So, placing constraints on how bots should behave around children may bring in to question the legitimate purposes of humans in general and I worry about closed mindedness in that debate. I'm hoping for some open ended solutions that can allow for adaptation to unpredictable changes in the environment.

It raised an alert for me, cause in conversations with friends and old coworkers most places seems to be actively silencing this

@munizdianamuniz I actually just released an experimental "green LLM" that attempts to resolve some of these inefficiency issues: https://github.com/johnmn3/mmllm