Yesterday's AI thoughts: manton.org

@manton Double entendre in the title intended? ;)

@manton Similar to my feelings toward your perception that “tech behaves badly” yesterday – I’d say try not to let yourself be held captive to the heckler’s veto.

Particularly when we are inundated with that narrative in a very biased media ecosystem, when actual opinion polling shows the public really likes tech companies more than most things in society and the products they build.

@manton No way, never stop blogging because it ruffles some feathers here and there. It’s your opinion, ppl should respect that.

@bax Thanks. “Heckler’s veto” is such a good phrase too.

@manton You always express your opinions thoughtfully. I am surprised and disappointed that these posts made people angry.

@johnbrayton Thank you.

@manton

@baldur had to stop listening to his podcast back in 2021 after he and his cohost repeatedly made clear their support for b-secamp

@manton

@sstephenson Ugh. I had completely missed that

@manton I think the reason the ScarJo voice situation is so unsettling is that we already know that OpenAI hoovered vast amounts of artistic works on the open web without giving credit or remuneration. I think we are all right to be skeptical of the company at this point.

@manton

@sstephenson @baldur yikes 😬

@cristian I agree that we should be skeptical and hold AI companies to a high standard. My posts were sort of a reaction to what looked like going from skepticism to getting out the pitchforks. 🙂 But you’re right that it played out this way because people already had negative feelings that were reinforced.

@manton I think I have read the post in question and it did not seem to be defending OpenAI or Sam Altman. And even if it was, and even if I disagreed with you, I would not have quit—or threatened to quit—micro.blog over it.

@manton I think yours is one of the rare balanced opinion for a person at the intersection of tech and blogging. People who really want to quit Micro.blog because of your AI opinions will do so without threatening to do so.

Recently at UT Austin, we had a 400-person webinar on UTA’s future direction in AI. They polled people on their initial impressions of AI/LLM, nearly 80% chose “curious” instead of a positive or negative opinion

@MitchW Thanks!

@pratik That is interesting about “curious”… Makes sense. This is all still so new, I don’t think most people have fully settled on how they think about it.

@manton Can you point out the places where OpenAI has done something to make you trust them so much?

@stupendousman Just in case you don’t know, search the plug-ins directory for “ChatGPT” and there are a couple options that should block it. I get that it does put faith in the robots.txt system, though.

@mcg Interesting question but is it really about trust? There are actual things that happened and we can draw conclusions from them. I do think OpenAI has created something very compelling, though… Whether it’s good or bad can be debated!

@manton We know how OpenAI got it’s training data. We know they disbanded internal safety groups. In the case of the Sky voice, we will never know the truth from OpenAI’s end, only some rather damning evidence of what was done in public and with Scarlett Johansson. The deference you’re showing towards OpenAI is worrying.

@pratik FWIW I don’t think the folks that want to leave are doing it just due to AI bad takes, but others in the past as well.

@mcg We might know the truth, which is why I posted that a lawsuit could actually be helpful. If there’s an internal email that says “interview 400 actors until we find one that sounds like Scarlett Johansson and then tweak the model to sound even more like her” that is going to leak.

@manton But it does sound like her, and they asked her to do the voice. Even if they didn’t already have a history of disregard to artists with training data.

@mcg There was a segment on CNN yesterday where they asked people on the street about the voice: some people said it sounded like her, some people said it didn’t. If I’m wrong, I’ll own it, but the way this story is being told isn’t quite right, which is what inspired my first post to begin with. The larger point is about narrative and reputation, not even the details.

@manton Have you listened to the voice? To me what matters is if the artist that was asked to do the voice, said no, thinks the voice sounds like them. Enough so, to lawyer up.

What part of the story isn’t being told quite right?

@manton

@baldur Those replies are not representative. If anything, Micro.blog probably leans more skeptical of AI than other platforms, because many of our customers have a healthy distrust of large platforms. Honestly being pro-AI or anti-AI is way too simplistic. This is complicated and far-reaching.

@mcg Let’s let this discussion drop for now and see how it plays out. That you would ask if I had listened to the voice before I wrote so much about it means we’re not on the same page. The truth is we probably agree on several aspects of this, and I appreciate that there is skepticism about AI. I have taken that to heart with every decision I’ve made about AI in Micro.blog.

@manton it’s clear OpenAI led by Sam Altman has next to no respect for creative work and predictably appear to have very little real creativity as well. We are living in the Age of AI no doubt and we are already using it whether we like it or not. Micro.blog has global settings and this is an excellent option

@mcg As is their choice. I mean, I left Twitter as soon as Musk bought it so who am I to judge?

@manton the AI deniers are just projecting their own fears and existential dread on Sam. They need a villain so they don’t have to face their own insecurities any an uncertain future. Much easier to project their avoidance on others than to face their own reticence.

@jhull any chance they get to avoid dealing with their own issues (like your blog), they’ll take so I wouldn’t take it personally. The deniers and “creative” crowd will likely do the same with my comments not knowing who I am or my background. Otherwise, they would see that they are the problem.

@manton i think you are dead wrong about AI. i believe that after all the lawsuits and hype have run their course, it will be all but non-existent. perhaps a tiny bit more useful than cryptocurrencies, but just barely. between your constant boosterism of AI, and your earlier claim that the world is “too woke” without being willing to define your terms: yes, i am pretty close to unsubscribing from your blog. and if i had an account at micro.blog, i would have already cancelled it. i also don’t read any republicans’ blogs, for similar idealogical reasons.

this is a thing i am allowed to do, despite your apparent belief that it is an overreaction.

@jhull Thanks. I wish it wasn’t so polarizing. When people take extreme views, it also hurts the legitimate arguments around safety that we should be paying attention to, and then we’re all worse off. I’m an optimist and want to build useful tools, not argue.

@manton

@acarson It’s super frustrating because micro blog is like the easiest gateways to the Indy web that I could locate fully managed. It’s super frustrating!

@manton

@crashglasshouses Probably both! I’ve recommended so many people to use that service because it was easy, and accessible, and it’s just super frustrating because I’ve directed non- technical users to that website and said, hey! This is a really great way to get started on the Indy web. I can’t express how frustrated and Angry I am