AI Excuses

I think I touched on it briefly in an earlier post, but if anyone has any doubts about my opinion I’ll make it clear: I think LLMs and AI in their current form are a plague.

I’m not going to write a long footnoted essay about the topic, as I don’t think I can say anything here that hasn’t been well covered by people like Ed Zitron or on excellent podcasts like Tech Won’t Save Us. I do have a particular thought, though, about the excuses a lot of otherwise well-meaning and generally thoughtful people make.

Nothing puts me off a tech podcast faster than blind enthusiasm or expressing the ‘inevitability’ of AI. I absolutely dropped This Week in Tech because they only seemed to have people on who agreed that the benefit of AI outweighed the damage it was doing to real creatives. Daily Tech News Show is generally better but still seems to have a lot of people that want to use it for everything, where the ’everything’ seems to mostly be podcast transcription or generating cover art for the latest episode of their podcast.

I can somewhat forgive the first part, because I think it’s a genuinely net-positive tool to have audio transcription for accessibility, and it’s certainly a time saver for lots of people that need notes and transcriptions. There are probably still ethical concerns on training material, as it was certainly from scraping broad swathes of YouTube and other places where audio would have accompanying subtitles.

The image generation is much less forgivable. Anyone who has ever made art and uploaded it somewhere has no doubt had it scraped and used without consent for training those models. Now anyone can come along and generate new images, cutting real artists out of the picture. If you bring this up, people inevitably whine, “but I’m not an artist, and without this I couldn’t have nice art for x, y, or z!” So learn to draw! Whatever you think of him, Pewdiepie did it. It took a lot of time and effort, but apparently he got pretty good. Don’t want to do that? Then hire someone. If you are doing something like a podcast and want nice art to give it a measure of respectability, you’ve got to do some math and figure out what that is worth to you.

What really got me going on this recently was an interview with Andrew Heaton on DTNS. The story was about him using generative AI to write a musical about the election. I don’t remember what prompted him to say it, but at one point he threw out all the excuses - I’m just one guy with a laptop, I only had a few months until the election, I don’t know how to write music, etc. That’s not an excuse! You can sit down and learn to write music, or maybe write some clever lyrics for songs and partner with a musician. Don’t have performers? Plenty of cities have community theaters that you could work with. People have managed before, we had musicals before we had AI.

I just don’t understand the excuse of “I want to do a thing, but I don’t know how, so I’m justified in using AI.” Not true! People just don’t want to do the hard stuff.

I’m also of the opinion that if you can’t be bothered to make it, I can’t be bothered to consume it. None of the content produced by AI is any good, because the people building those tools don’t want to consume the output either. They want to sell it to someone else so they can make a quick buck. The people buying it want to replace real people with mediocre output so they can make a quick buck. And the state of the internet is very reflective of that right now.

Today’s header - I got a Canon R10 a while back so I could learn some photography skills. I got this bee while playing with a macro lens in the garden.

• Ai