Discover more from Scenes with Simon
Scenes with Simon #3 - The Many-Fingered Conundrums on Detecting & Banning AI
Banning AI, Grand Central Madison, & Power Dynamics in White Lotus
The Many-Fingered Conundrums on Detecting & Banning AI
Medium announced this week that articles written by AI must be labeled as such. It will not be up for remuneration from Medium’s business model and it won’t be distributed on the network. Other writing platforms are more stringent.
Any aspiring writers who submits AI content will be barred at the gates like the uncivilized barbarians they are.
Any current Fanfare writer that submits an AI-generated story will likewise be shown the exit. No second chances, no redos. That may sound harsh but I want to be completely clear on this.
How do we detect AI works if we can’t trust someone to label it as such? And what do we risk in doing so?
Currently, AI generated works still has some of a visual/gut test. For images, just look at the fingers, they say. That’s the clue to it being AI generated or not.
via MidJourney: “a realistic image of a young man typing a newsletter on a keyboard”
I have no doubt the visual finger test will eventually fall away as the models improve. But, what if it’s text (generated by ChatGPT)? There are detectors out there that seemed to be able to discern whether it’s AI generated or not.
After running the opening paragraph of my debut novel through these, I believe that I am human. One of them was a bit suspicious, believing that I’m 0.03% fake. I’ll take those odds.
Besides these detectors employing various techniques, some of it do slip through. That is, unless AI generators start employing watermarks, hiding a fingerprint in the generated text that would require a lot of wrangling to fully remove. Will a cat and mouse game develop? For sure. Especially as open source models might strip away watermarking as a matter of principle.
I understand why there’s a desire to ban AI works. Besides the discontent around how AI impacts power dynamics: not remunerating training data (whether it’s legal or not) + techlash around a few people making billions from automation at cost to the creative working class, I believe there’s also a different, unexamined fear. It is the fear of truly enjoying works created by AI (even if it’s manned with intent by a person). It crosses a threshold of automation that makes us feel that intent (and the artist “struggle” had been lost).
The dark forest of generative AI is the fear that we wouldn’t be able to judge the realness in the online.
After the forest expands, we will become deeply sceptical of one another's realness. Every time you find a new favourite blog or Twitter account or Tiktok personality online, you'll have to ask: Is this really a whole human with a rich and complex life like mine? Is there a being on the other end of this web interface I can form a relationship with?
Before you continue, pause and consider: How would you prove you're not a language model generating predictive text? What special human tricks can you do that a language model can't?
But, if we formalise the detection and banning of AI, we run the risk of having our (five-fingered) hands stuck in a fingertrap toy. Resist and we will be captured by the unintended consequence of the fear of judging and being judged to be human.
If the automated systems to detect AI fail, but we believe it to be inhuman, is it our place to judge the humanity of the work? Who, after all, wants their work to be judged to be inhuman?
On the other hand, we can choose to let AI flourish as equals. But then, we also run the risk of losing the sacred act of every creative to suck and to be judged as among humans. To not live in a hostile dark forest. To allow creatives to start with stick figures, poor grammar, and a full heart. To be real. To be human. To be separate from the machine. To suck.
This, from Paul Graham, I believe, is dangerous:
I say, write that essay even if ChatGPT could write it. Even if it’s “conventional”. We want that process to remain and in some sense, we want to keep that sacred. Start, even if the AI might be better than you.
And therein lies the conundrum.
By banning AI, we ensure that we don’t compete with the AI. We ensure that people can start. But, we also elevate the AI and run the risk of being judged to be so “pedestrian”, so mid, so run-of-the-mill, so inhuman. I don’t want an automated or human detector running across my work at every gate of the public web. The NPC memes are awful as it is, and now there’s more ammo online to be judged to be inhuman.
By fearing AI, resorting to detecting and banning it, we indirectly serve judgement on ourselves. We say that we are inadequate. Yes, we want to talk to humans, not bots. Yes, and we also want to experience works made by humans over AI, but in creating this distinction, we judge the works of others, not based on its merits, but whether it’s human or inhuman. And that, feels dangerous. In some way, I say, bring on the AI and let’s not be afraid of it.
We want people to use AI to imagine and create before they are able to be better than it. One’s imagination will always outpace one’s ability to create, after all. We want to see those niche stories of a sci-fi loving teenager from Guinea with a big imagination. Let’s maybe reserve the judgement and rather just try to enjoy it before we fall into this trap?
I don’t know the right answer, and this conundrum isn’t the only part that matters of the AI creativity debate. Judgement of works is but one part of it. Who knows where this rollercoaster will go?
Remember to keep your many-fingered hands inside the vehicle at all times and enjoy the ride!
Filed Under: communities, online
Speaking of touching grass.
I found that this post from Noah had a great framing of communities vs ones we are used to. Not online vs local. But vertical vs horizontal.
Vertical communities always existed. It's just now more prevalent. This ties in well wrt to the dark forest of AI. I also think that the collapse of horizontal communities (especially in places where there are poor 'third places', like car cultures) has in part contributed to social upheaval.
We're also not that well versed in understanding how to treat vertical communities. We treat them like horizontal communities. Threats are the same: whether in our ‘real villages’ or our ‘filter villages’.
In vertical communities, we don't have access to as much common ground as a face to face with another human. So, it leads to this inexorable further retreat into building walls around our vertical (and horizontal!) communities. In other words, if there was something that I believe is important, is that we should reinvest in physical communities: increasing funding for local businesses, groups, culture, sport, etc. Touch grass with more people. In some way, it reminds me of the importance of conferences in the crypto industry: a primarily vertical community that touch base in physical spaces every so often. I’m reminded of the writer and counselor Esther Perel that said: not everyone needs sex, but everyone needs touch.
Filed Under: trains, urbanism, transport
After 14/15 years of construction and roughly ~$11B later, the Long Island Railroad in New York can now go into Grand Central Station instead of Penn Station. This is a great thread showing what it looks like. This link was done so that commuters didn't have to transfer between Penn Station and Grand Central (for whatever reason). It’s the biggest new rail terminal in the US since the 1950 & has 8 miles of tunneling while not shutting down any service at Grand Central. The B1M has a really great overview of the project.
Speaking of new underground terminals:
Filed Under: urbanism, bicycles
Oh, imagine all cities having infrastructure like this? Safe bike lanes and parking for an extra 7000 bicycles opening up under the central station?! Here’s a 4-year timelapse of it being built (they had to expand and drain a part of the river nearby).
Filed Under: urbanism, malls
If malls are to continue to exist, I honestly feel that, the faker the better. I love these heterotopic, weird, third spaces. In a way, like Meow Wolf's Omega Mart. Not necessarily a Disney-ficiation, but a fever dream-ization. Lean into it.
That, or like as commenter wmmseo mentions: malls should come combined with residentail built-in. It's bizarre that we're still thinking of zoning as something that has to be separate. Have suburbs that can have small mom & pop retail in a garage AND we can have mega mall with residential + offices also built into it.
Filed Under: karnivool, prog, metal, music
Karnivool's "Sound Awake" is one of my favourite albums of all time. Just a glorious progressive rock/metal album throughout. This live rendition of "Change", 10 years later, is amazing. 14min of beauty. Put it on, sit back, and take it in.
Filed Under: fantasy, writing, magic
In December there was quite a thread on Twitter detailing "unpopular" opinions about fantasy. Matthew draws an interesting distinction about the purpose of a magic system to the story.
Filed Under: finiliar, nft, dynamic
I love these cute NFTs. They are dynamic and change based on price action of certain assets. :)
Filed Under: japan
Another article from Noah in this newsletter, detailing ways in which Japan has changed since the 90s.
Filed Under: music, guinea, vibes
I love Ghost Culture, and his production work here with Guinean musician, Falle Nioke is so good. Such a strong drive to this song.
Filed Under: nft, art, receipt, dynamic
Big fan of this art project. I’ve been wanting to create a similar NFT for those who had held “This Artwork Is Always On Sale”.
It’s all on-chain SVG too and automatically records its own transfers. And those textures? So realistic. Just all round hearts for this one. <3. (ht ChainLeftist)
Filed Under: music, naming
One of the most interesting question for a musician is: if you want to experiment and reinvent yourself, do you keep your brand or start with something new? Some bands have become renowned for reinvention while in other cases, fans dislike it when the musician changes. In some cases, musicians change their names when they do different styles and projects. I actually don't know what the right answer is? Why does this matter? Avalon Emerson is doing something entirely new, a lot more chill and also vocal. Very different to what she has made before, and so this question came back to me.
Filed Under: cities
Yanjin is by far one of the most interesting cities, nestled alongside mountains and a river in China.
Filed Under: robot
Watching robots fail is sometimes more interesting because it feels more human. (ht Fred)
Filed Under: flying, nasa, boeing
You might have seen posts about this new NASA + Boeing sustainable aircraft design with the long wing + truss design. This is a great video describing why this approach was taken.
Filed Under: whitelotus, power
I’m a huge fan of White Lotus. Season 2, I feel is an instant classic that will be studied for years. This video essay captures what I enjoyed about the show: how well it captures power dynamics in different ways. It’s subtle, but it also feels complete and perfect in that sense.
Filed Under: dnd, creativecommons, IP
Wizards of the Coast eventually just lets up and says that they won’t be rescinding the OGL anymore and that the core rules for the 5th edition is now placed under one of the most permissive creative commons licenses. I’m honestly tired of this story. They screwed up. They started backtracking. Then they backtracked even more to a point where the licenses are now more permissive than when they started, but they still lost trust from the community. Will take a while to rebuild that trust. I still think a part of this decision has to do with the film coming out soon. Step back, let the dust settle, hopefully have the film succeed, then more methodically re-open conversation for the new DND edition.
Although Wizards of the Coast was caught with their many-fingered hands inside the greedy cookie jar, big brother Hasbro will continue to breath their dragon fire down their necks.
Thanks for reading friends.
Enjoy the sunset.
Subscribe for free to receive new posts and support my work.