I do kinda a gree with the take though. I think most flaws we see with AI are just flaws in society exacerbated by AI: Greed, inequality, laziness, disregard for the impact of your actions on others, ...
As someone with few artistic talents I am able to use AI to help share my imagination. It’s not perfect, but so many artists are gatekeeping because they fear their livelihoods and wellbeing. The emotion is real. The root though is societal structure.
Art is important to society and provides tremendous value. AI helping create it doesn’t lower its value. If anything it gives voice to the previous voiceless artist. It may result however in significantly more “bad” art, if there is such a thing.
As an AI researcher in academia, industry has super botched releasing these models to the public.
(And made the science harder too by making reproducing experiments almost impossible)
Pretty much, yeah. But also they tend to change the model without telling us that it changed or how it changed. We have to put exactly what version we're using but we have no idea how gpt-4o-2024-08-06 differs from gpt-4o-2024-05-13
I think the negative social effects is more a subject of how we use it. Powerful tools have equal capability to improve or destroy lives. Similarly, the ecological impacts can be massively mitigated by improvements in energy efficiency.
I think the issue we have is abuse of tech by technocrats
If course it will if we let them, as with any technology
Neither ignoring it or demonizing the technology outright is going to help us, the proverbial genie is out of the bottle and we can all see the amount of money flowing to it.
What we need to be discussing is guiderails and regulation
For sure! And of course, AI existed way before this big "boom" of it in many forms. There isn't a way to avoid it completely. But we can definitely be thoughtful about what kinds of AI we engage with and making sure that we're using it for good reasons, +
not "lol I wanna see a cat with a sword but I don't want to pay an artist" or "I don't know what to write in this thank-you note" and DEFINITELY not "I want to fire half my staff and automate their jobs, all while putting out a lesser quality product and raising the price for consumers"
And I can recognize that I don't have the tools or knowledge base to do much to make improvements in energy efficiency. What I can do is ignore as much of the AI being thrown at me as I can. If most people did that, the technocrats wouldn't be inclined to invest in AI
I guess what I'm trying to say is the convo we should be having as a society is how best to socialize the benefits of AI and mitigate the human and ecological impacts, and hold technocrats accountable to that.
I want to be clear I have a very firm stance on AI as a tool to aid humans especially in the areas of HCI. Eg a phenomenal use of AI is in screen readers to generate image descriptions for visually impaired people.
The problem we have is a social one, driven by greed, to devalue humans
Thank you as someone who uses a screen reader part time, people confuse AI and ChatGPT to be the same thing too often. I have several chronic illnesses that prevent me from accessing technology normally and AI is necessary for me to use things like VoiceOver, Voice Control, and eye tracking
I can agree with these things! Use of AI to assist with disability is great! As long as we don't forget to continue to educate able-bodied people and make sure we remain inclusive.
I fear if we start setting expectations that AI will solve those problems, it will become the disabled person's responsibility to maintain access to those tools, which would likely turn it into a market, and allow for the already privileged to remain ignorant.
Absolutely! Even in this example im not arguing that AI should be used to magic wand, far from it. Treating it that way is a huge part imo of how we're "abusing" it.
There's no true substitute for actual, thoughtful accessibility features, same for actual people. AI should not be a replacement
It's a tool, and I think in most, if not all applications, it's always better if there's still a human in the loop. Yes we can have AI generate alt text, but it's always gonna misrecognize stuff, or be overly descriptive and miss the point of the image in context of the content where it's posted.
But broadening out AI from the narrow context of LLMs the current hype revolves around: we've had deep learning models now for some years that have been way more accurate at detecting certain kinds of cancer from scanned images than humans, making it an amazing tool in the hands fo a trained doctor.
Similarly, AlphaFold has made huge strides in understanding the structure of proteins, which used to take scientists years, and lots of funding, so it's a huge advancement in medical science. I feel like all of these aspects of AI are being overshadowed by the current hype
Comments
(And made the science harder too by making reproducing experiments almost impossible)
I think the issue we have is abuse of tech by technocrats
Neither ignoring it or demonizing the technology outright is going to help us, the proverbial genie is out of the bottle and we can all see the amount of money flowing to it.
What we need to be discussing is guiderails and regulation
That said, there are hundreds of companies actively replacing human labor with cheap AIs. Call centers, support positions, communicators.
We need to act on that
They see it, philosophically, as "the next step" for capitalism, as a way to replace "human capital" with unpaid AI that doesn't want health benefits.
Plus, you can advocate for renewable energies!
I guess what I'm trying to say is the convo we should be having as a society is how best to socialize the benefits of AI and mitigate the human and ecological impacts, and hold technocrats accountable to that.
Much easier said than done, i admit
The problem we have is a social one, driven by greed, to devalue humans
There's no true substitute for actual, thoughtful accessibility features, same for actual people. AI should not be a replacement