Post ASbWiFpMDmZHw6Yo9w by TedUnderwood@sigmoid.social
 (DIR) More posts by TedUnderwood@sigmoid.social
 (DIR) Post #ASaKV81uVYARNcRlK4 by afamiglietti79@mastodon.social
       2023-02-11T23:10:10Z
       
       0 likes, 1 repeats
       
       This is the most coherent version of the argument I've been trying to have with @TedUnderwood  today. The TL;DR version is: "What social benefits will we deny people if LLMs are delayed in their mainstream adoption for a bit? Shouldn’t there be at least some affirmative duty to make that case before we push this out to most of humanity like a software patch?" https://afamiglietti.org/uncategorized/please-stop-moving-fast-and-breaking-things-ai-edition/
       
 (DIR) Post #ASbUC5mPqHL9bEikNM by ctbk@mastodon.uno
       2023-02-12T09:17:31Z
       
       0 likes, 0 repeats
       
       @afamiglietti79 @TedUnderwood OC the rush is the fear that, while you pause to ponder, another competitor with even less moral principles than you will rush and steal away your precious future market share that you *could* have grabbed if you just went ahead, caution be damned. In other words: how to put ethics in business?Even harder: how to do it w/o stifling innovation? With law? Having a sort of worldwide tech council entity that regulates the industry? I don’t know.
       
 (DIR) Post #ASbUC6Ndbt79Sg8TOS by TedUnderwood@sigmoid.social
       2023-02-12T12:44:13Z
       
       0 likes, 0 repeats
       
       @ctbk @afamiglietti79 and while “Let’s just slow down” sounds appealing, it’s also not really clear that we would have more  consensus about the social effects of AI two or four years from now than we do today. We’re not talking about a new drug where we could do controlled studies of its effect before release; this is a social transformation that is necessarily controversial and hard to predict.
       
 (DIR) Post #ASbUC7ocH4HxudlX5E by ctbk@mastodon.uno
       2023-02-12T09:24:25Z
       
       0 likes, 0 repeats
       
       @afamiglietti79 @TedUnderwood the underlying issue is the impossibility to held companies accountable of the social issues they generate/exacerbate with their profit laser focused behaviors/products/algorithms. Look at the ridiculous (lack of) consequences Big Social had to face after the various messes it did all around the world. At most some little slap on their hands, when there was anything at all.
       
 (DIR) Post #ASbUnZfLgoZF0JHv0q by afamiglietti79@mastodon.social
       2023-02-12T12:50:59Z
       
       0 likes, 0 repeats
       
       @TedUnderwood @ctbk sure, but that example reiterates my point. Delaying a drug has costs, sick people aren't treated. What's the cost if these are delayed? At the very least we could run the experiment where we see how influential an AI brainstorming companion is a few more times, and explore the INFINITE amount of writing they can produce enough to rough out what percentage of it tells people to throw batteries in the ocean or whatever.
       
 (DIR) Post #ASbVHzUwD8rroAvEcy by TedUnderwood@sigmoid.social
       2023-02-12T12:56:27Z
       
       0 likes, 0 repeats
       
       @afamiglietti79 @ctbk The good thing about human liberty is we don’t have to prove our actions will have a known good effect before taking them; we’re allowed to do stuff just because we want to, or because we want to find out what will happen. The burden of proof is on the other side.
       
 (DIR) Post #ASbVkCeIuQaKhhkuY4 by afamiglietti79@mastodon.social
       2023-02-12T13:01:35Z
       
       0 likes, 0 repeats
       
       @TedUnderwood @ctbk that's not true for drugs, it's not true for all sorts of stuff. The more I think of it, the more I think Plato was onto something, treat it like a drug...
       
 (DIR) Post #ASbW0HVabr0wPOLWsq by afamiglietti79@mastodon.social
       2023-02-12T13:04:30Z
       
       0 likes, 0 repeats
       
       @TedUnderwood @ctbk and we don't use the principal of individual human liberty to guide what companies can do when they roll out products to the world, in almost any instance.
       
 (DIR) Post #ASbWiFpMDmZHw6Yo9w by TedUnderwood@sigmoid.social
       2023-02-12T13:12:24Z
       
       0 likes, 0 repeats
       
       @afamiglietti79 @ctbk Drugs, motor vehicles and food safety we do regulate. But you can make furniture any color you want, you can put any words you want in your novel or magazine, and print thousands of copies. You don’t have to prove there’s a compelling social reason for fidget spinners to exist, or that the known god effects outweigh the bad.
       
 (DIR) Post #ASbXKSGARj2PUzJei0 by afamiglietti79@mastodon.social
       2023-02-12T13:19:21Z
       
       0 likes, 0 repeats
       
       @TedUnderwood @ctbk do you really think that dropping ChatGPT into Bing is no more consequential than IKEA releasing a blue bookshelf (and they can make the shelf whatever color, so long as the pigment is non-toxic, true, but if they change the weight distribution without one of those toddler tie-downs that's a Whole Thing). I don't think a particular piece of writing is equivalent to an entire technology of writing.