thenewleafjournal.commentioned
In March 2024, I published an essay titled Humane Content in World of Language Models. While I have since taken a stronger stance about describing meaningful writing and media as content, it fits here. This essay was prompted by an incisive essay by Maggie Appleton titled The Exp
gpottermentioned in href weekly – Jan 1-7 2023
Bits https://flipperzero.one/ Flipper Zero is a portable multi-tool for pentesters and geeks in a toy-like body. It loves hacking digital stuff, such as radio protocols, access control systems, hardware and more. It’s fully open-source and customizable, so you can extend it in
@hdv this is insightful and clear as always, but I wonder if she didn't dismiss certification too quickly. E.g. I remember the Thawte-backed Web of Trust fondly, which was before its time in decentralizing identity. (And I'm sad to see @maggie seems to have abandoned the fedivers
@sil interesting read and unfortunately fast becoming a reality. I call it the insta web. 'I want it now, I don't care about tomorrow or the path of destruction I leave'. I find myself spending more time surfing Gemini spaces than the web these days. Maybe the refugees of the web
@hdv there may be a level of originality, but that’s a fool’s game to discuss; it leads into a bunch of philosophy of mind stuff about “what is creativity” and the like, which is exactly the discussion they want us to be having because just talking about that stuff confers legiti
@hdv @maggie jeez Louise, this is depressing. I mean, I agree with it entirely (and I hadn’t heard of the dark forest metaphor, which is alarmingly evocative), and it’s good to lay this all out in a way that will open people’s eyes as it did mine.This is only kicking the can down
@sil as techies we could publish RSS only so that it is too niche to be discovered by those training models… no use for the wider world though. I also wonder, hopefully, if there is a level of originality that these systems can never emulate? Surely there must be a limit to this!
I went to Dan Kay's original (where it was posted pre-Medium) to check out his blog. The last comment there linked me to this. Wowza! And the mutual follows tell me I "should" already know Maggie Appleton, maybe even from G+. #serendipity #curiosity
"How would you prove you're not a language model generating predictive text?" Year 2030. You meet a girl at a bar. After chatting for a while, you both look each other in the eye and say the most terrible slurs you can think of. She's legit.
DavidNEmentioned in Jungles, Gardens, and Hope
The ability of humans to have hope is a remarkable thing; we live in a world where probabilities insist on things like suffering, poverty, and tragedy. It is a world where we bicker endlessly from early childhood until old age. And yet we have hope, it is fundamentally human to h
A reminder that what we see online is already curated by bots & AI. Social networks & popular portals to the web are using them to place & promote content. As we consider the future, let's be mindful of the AI's power to shape our reality in the present
The AI boom is equally exciting and scary. "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 you'll have to ask: Is this really a human with a life like mine?"
Are we slowly building a internet devoid of humans? For bots by bots - where our unique humanness is pushed into exile? "Our new challenge as little snowflake humans will be to prove we aren't language models. It's the reverse turing test" - @Mappletons
piersstobbsmentioned in February Newsletter
Hi everyone- January reminded us (in the UK at least) of the joys of a ‘big coat’ with 2023 definitely off to a cold start… No great change to the depressing headlines though so hopefully time for a bit of distraction with a wrap up of data science developments in the last mon
Veo cada vez más personas en twitter limitando sus cuentas y usando modo privado porque es una forma de protegerse de los bots y de un espacio hostil. Recomiendo la lectura de esta nota respecto a eso: via @dtirado
datamachina.commentioned in Data Machina #183
AI Pair-programming. Code prompting. Neuroscience + neural nets. Graph transformers. Neural search. Neural collaborative filtering. Diffusion GANs. Google Muse. MS Research VALL_E. On AI Pair-Programming, Code Prompting & LMs. I’ve been doing research for a company on all this
cybe.pressmentioned in I Make no longer Write
I in actuality believe currently been made conscious that the frequency of contemporary voice published on my space has long gone down slightly a small bit.Good ample swish, I trash-talked Manuel’s internet space on Mastodon and he accurately identified that while I wrote a ambit
How to combine "Artificial Intelligence" with "Human Intelligence"? Incorporate distinctly human elements like reasoning, feelings and emotions into a human intelligence platform -- for example: app.pollthepeople.app/results/3033/8…
interesting essay on ML bots and how they may impact digital life, but what I’m really loving is the ecology metaphor for the digital ecosystem. ML bots as mushrooms and what not on the forest floor digesting digital content, for what?
This is both a great read and has one of the best pictures of web structure I've seen in very many years. It will be a long time before AI can do this! cc: @charlesarthur
Warren Ellismentioned in The Garden Is Frozen
It’s hard to write a lasting story on ice. Recurring cold snaps and hard frosts have frozen the garden in place, and it’s impossible to work with. I ordered a bunch of dormant plants, as this is the season to pick them up cheap, and now I’m just hoping they survive until spring
This is an important article. "We're about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft."
Più che futuro, mi sembra tutto passato, ma in questa lettura ci sono riflessioni più approfondite e divertenti sull'interazione uomo-web, che poi di questo parliamo. Da vent'anni. #ArtificialIntelligence
manualdousuario.netmentioned in Meu trabalho em risco
Alguns séculos depois, sinto hoje o que artesãos e pequenos produtores ingleses devem ter sentido quando viram chegar as primeiras máquinas e serem inauguradas as primeiras fábricas durante a Revolução Industrial. Tecnologia recente, as inteligências artificiais (IA) gerativas re
Digitalia.fm - Backchannel mentioned
The Expanding Dark Forest and Generative AI Assumed Audience People who have heard of GPT-3 / ChatGPT, and are vaguely following theadvances in machine learning, large language models, and image… January 20, 2023 at 12:25PM via Instapaper
1. Interesting 📰 about the incoming exponential increase of the web’s dark forest (dead zones w/ no human input/presence) due to ai & how in the future most virtual social spaces will be overran by bots (wayyy more than it is now)
And what about the Expanding #DarkForest? "It's like a dark forest that seems eerily devoid of human life-all the living creatures are hidden beneath the ground or up in trees.If they reveal themselves, they risk being attacked by automated predators"#AI
stephaniewalter.designmentioned in Pixels of the Week – January 15, 2023
On Twitter and LinkedIn and Mastodon, I share curated articles I read, resources and tools about UX Design, User Research, UI and mobile design, HTML, CSS, the web industry, some process, some inspiration, etc. This is an archive of everything I shared this week. And some extra l
ICYMI "As consumers and creators, we'll [need more] critical thinking and skepticism...In a world of automated intelligence, goalposts for intelligence will shift...raise our quality bar for what we expect from humans" #LLM
"Our new challenge as little snowflake humans will be to prove we aren't language models. (...) Would it even be the web if everyone knew you were really a dog?" Good read about the need to prove you're human on a web flooded with generative AI content.
Don't mind me, just adding a stroke of horror to your Friday the 13th…🩸 "Dead Internet Theory posits that most of the internet is empty and devoid of people and has been taken over by artificial intelligence."
#AI An essay on the gloomy forest of growing AI generated content, and the needs and means to prove you are human, and stay human, in the global mess of generative content:
"[Language models] don't have vision, sound, taste, or touch." "They don't know about events that happened after a certain cutoff date." "In short, they do not have access to the same shared reality we do." Good read 👀 #ai #llm #chatgpt
Ben Jones | @benjones@mstdn.social mentioned
My point, expressed much better by @Mappletons: "We're about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft."
"We’re about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft."
Useful article with a great illustration on humans vs. chatGPT et al. and where to find real as the AI infiltrates. @kyleracki has a great article on leveraging AI to plan that real content (video in his case): linkedin.com/feed/update/ur…
This article is a great read which I think has tangential implications for teacher credibility and authenticity with regards to AI generated resources. I can see AI resources flooding TES and edublogs.
Internet will soon be so flooded with #AI-generated content that original, human-written content will be unavailable. Do we need to invent a new Internet only for humans, or to come back to real life relationships? #Opinion
I don’t even dislike the article, I just don’t think this is anything new or that ai is just gonna change the digital landscape or culture downstream, just that it’s already in flux/crisis.
FWIW, Facebook knows who the real humans are. Or to put it a slightly different way, I know who the real humans are in my feed... I do see a touchpoint of reality hitting a blockchain of authenticity for verification at some point.
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?
Any Thoughts about this #DarkForest of the #WEB that is forming by #AI Are we going to have to keep being more & more original to avoid being targeted by #algorhythm of Machines?
"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?"
Danielle Lee Tomson PhD mentioned
"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?" Y'all been NPCs so long that algorithms sound more human. h/t @acookiecrumbles
"The Dark Forest theory of the web points to the increasingly life-like but life-less state of being online. Most open & publicly available spaces on web are overrun with bots,advertisers,trolls,data scrapers, clickbait & algorithmically manipulated junk"
#DarkForest: "the life-like but life-less state of being online: Most open/publicly available spaces on the web are overrun with bots, advertisers, trolls, data scrapers, clickbait, keyword-stuffing “content creators,” and algorithmically manipulated junk"
Loved this @Mappletons. "The Reverse Turing Test" made me want to jump out of my seat when I read it. I was reminded of that scene in Ex Machina where the engineer cuts himself to make sure he still bleeds to prove to himself he's human.
What about the halting problem? Can you prove your humanity by reasoning about code to decide if a given program will terminate? I guess it depends if a model can abstract away it’s turing machine-ness enough to get past that limitation.
Enjoyed the read! Really feel there's still space for human-produced content on the web. But GPT will definitely be the #1 source for mass market content
Awesome post. I’m not too worried myself. As an educator I can recognise when students have their own voice. You yourself, Maggie, have one that I recognise! So I look writing with a ‘voice’. AI may develop to have a voice- but if indistinguishable & inspires me, it won’t matter
It’s also true with AI art. If I saw a picture that moved me, that would be a reason for me to buy the art or put on my wall. If I read a book that was amazing and turned out AI wrote it, it wouldn’t put me off. It’s the journey/experience that matters.
Maggie Appleton - indieweb.social/@maggie replied
Oh god that is horrifying. I didn’t expect this piece to take off like it did; there’s so much more to be said about possible repercussions. Lobbyist LLMs are viscerally scary. Similarly we could have patent troll LLMs, litigation LLMs. Just a whole expansive suite of conmen LLMs
older-/wiser-me feels such a word is becoming increasingly important: 'trade-off'. it's almost like our whole-life is a continuous trade-off to navigate. "But we'll also need to reckon with the trade-offs [...]"
we've seen a flood of LLM copywriting and content-generation products come out: Jasper, Moonbeam, Copyai, Anyword are just a few. They're designed to pump out advertising copy, blog posts, emails, social media updates, and marketing pages #chatgpt
Have you heard about the Dark Forest Theory 🌲 of the Internet? It suggests that the web is becoming increasingly devoid of genuine human presence, overrun by bots, advertisers, and algorithmically manipulated content. | @Mappletons
"We're about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft." The Expanding Dark Forest and Generative AI
Freerk (@freerk@zirk.us) mentioned
„We're about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft.“ – by @Mappletons
Lars "Land is a Big Deal" Doucet replied
What does he think of these two articles: fortressofdoors.com/ai-markets-for… (RE the potential for generative AI to change the signal to noise ratio of human versus bot generated content given decreasing marginal cost of automated content creation)
In a world flooded with AI copy, "... [r]eferencing obscure concepts, friends who are not famous, niche interests and recent events make you plausibly human. This feels eerily like a hostage holding up yesterday's newspaper to prove they are still alive."
"We're about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft."
"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?” – Questions I never thought I'd ask myself 5 years ago
I think people will pay to teach AI about their products and there will be laws about product placement in AI results, like paid spokesperson laws. AI will create videos across SM platforms and kids will listen to them over their parents because they are the performers.
When we hit the lower levels of AGI the web will mostly disappear. Why write a web page when GPT's answerbot is where everyone will go for info. (Who would use Google? It's like using a card catalog at a library. Just ask an expert.)
“You thought the first page of Google was bunk before? You haven't seen Google where SEO optimizer bros pump out billions of perfectly coherent but predictably dull informational articles for every longtail keyword combination under the sun.”
Although I think there’s an even milder version of this theory that’s just as pernicious: that the constant dopamine of the “feed” has turned many of the real humans into roving internet zombies with little beyond hive-mind hot takes to contribute
Hacker News Updates ⚡ mentioned
Microsoft is preparing to add ChatGPT to Bing → bloomberg.com/news/articles/… The Expanding Dark Forest and Generative AI → Intelligence – A good collection of great OSINT Resources → github.com/ARPSyndicate/a…
Maggie Appleton - indieweb.social/@maggie replied
In this optimising-for-connection world it’s super important to know if the “person” you’re talking to is a full, complex human you could grab a drink with, swap personal stories with, bond with, and generally have a longterm intellectual connection with. LLMs aren’t there yet.
'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 online, you'll have to ask: Is this really a whole human with a rich and complex life like mine?'
Maggie Appleton - indieweb.social/@maggie replied
It depends why you’re writing and publishing on the web. I do it to make friends and form relationships with IRL people. Most of the people in my community do the same. Which works out! Fulfilling intellectual bonding for all! Optimising for connection, not disembodied insights.
The Expanding Dark Forest and Generative AI #text #human #ai #generative #models #dark #original #language #forest #expanding #web #model #sense #world #dont
I'm not interested in 'the internet is x% bots'. Does it matter if we don't experience it? More interested in - 'what % of material you read is bot-generated?' Got any stats?
____.,,^\^..^/^,,.____ replied
This should only be growing pains though. As the generative AIs get better and better we'll doubtless come to prefer spending significant time with them rather than mere humans. The movie Her showed the way IMO.
Thanks for the thought provoking piece! Another thought - with content production becoming so cheap, human attention becomes even more valuable -> human curated digital gardens can be valuable digital gathering places + "beacons of humanity"
Looks like the robots are taking over the content production game. Don't worry, I'm sure they'll make sure to include plenty of emojis and awkward attempts at humor to let us know they're not completely soulless. 🤪#robotrevolution #humorattemptfail
On second thought, maybe we started with the wrong premise. The web is dark due to bots, trolls, and people spreading fake news and manipulating opinions, as well as criminal activities.
I feel like we’ve been actively stoking this through marketing, public relations, ads, sponsored content, etc. Creating an environment where trust is impossible, all interaction is a sell, breaking down peoples critical thinking, and creating a wealth of manipulating source data.
no it can't Google is a propaganda machine now, search happy White woman dead internet theory is old news bots already do most copy, reviews, likes, posts same in the market, 95% of trades are bots, has nothing to do w products or services all fake futurism.com/the-byte/co
4E cognition (embodied, enacted, embedded, extended) would currently be the main distinction. Temporarily. AI will make progress by becoming agents in environments. Feelings and emotions, human nature such as egoism, greed, desire, and loving-kindness are further away.
The problem/opportunity is that AI is becoming more literate than average humans. The issue is not whether the source is human, but whether AI adds value to the human using the tool. If Afgan women cannot attend school...do they care if AI is not human?
This was a super interesting article! Although I'd argue that something desperately needs to change with places like google. I think things like GPT3 seem so revolutionary because places like google have been stuck in an SEO cycle, making it impossible to find digital gardens.
CuddlySalmon | nptacek.eth replied
this was excellent! one thought on point 3 (ingroup language as a firewall) — if i can map my network graph like this, so can AI. what stops hyper-focused training of ai models on these clusters rather than the network as a whole? i could see that as a potential foil here
I dev/manage a few corporate websites and the flood of spam emails offering GPT-generated "guest posts" where the only original thing is a sponsored backlink is just abysmal. 50+ emails daily.
dave, aspiring peasant mentioned
"We're about to drown in a sea of pedestrian takes. An explosion of noise that will drown out any signal. Goodbye to finding original human insights or authentic connections under that pile of cruft." @Mappletons says a lot of what I've been thinking:
Yeah the arms race scenario seems like a LOT of compute power and energy ultimately pointed at nothing much… but it seems unavoidable to some degree It’s gonna be a bumpy ride!
“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?”
Buen análisis sobre el nuevo escenario de la web, llena de ruido generado por inteligencia artificial, bots, clickbait, trolls y basura manipulada algorítmicamente
"In a world of automated intelligence, our goalposts for intelligence will shift." - @Mappletons on memes as a torchlight for humanity under the looming canopy of AI generative-genericism
The dark forest theory of the web points to the increasingly life-like but life-less state of being online. web are overrun with bots, advertisers, trolls, data scrapers, clickbait, keyword-stuffing “content creators,” and algorithmically manipulated junk.
The Expanding Dark Forest and Generative AI: easily capable of outposting every human on earth, ChatGPT and LLMs like it will flood the web with a neverending tsunami of SEO bullshit
That makes me happy! This is tricky to game out. I'd love to chat at some point, about the work you do at ought, and some anthro perhaps, how that can help us navigate this labyrinth (anthro is my background too).
🔗 The Expanding Dark Forest and Generative AI - The web will be full of (good enough) text from the AIs. To maintain your human edge, be more weird (or “unique”) and talk with other people IRL more.
"Most open and publicly available spaces on the web are overrun with bots, advertisers, trolls, data scrapers, clickbait, keyword-stuffing “content creators,” and algorithmically manipulated junk."
@wfm I thought so too, very much, which is of course why I shared it. I’m kinda glad that my bookmarks in my Known stream end up here as direct links to the thing itself. Saves you a click.
@JMaxB @nporter Maggie Appleton writes super smart stuff; even when it isn’t directly relevant to me, I tend to read it. And this was super relevant. I’ve recently had an avalanche of spam offering AI to write for me.
This was great! I had my come-to-jesus moment in april, and shifted how I approached writing in response. For now, that has been a good thing, having this humanness-forcing-function. But it seems LLMs etc will increasingly break public meeting spaces, circumventing defenses. So.
Maggie Appleton - indieweb.social/@maggie replied
Yup, true. Although I think vastness is less of a problem than economics. Attention = ad $$$. Lots of people incentivised to make low-cost click bait. Hard to find good stuff when you’re drowning in rubbish. But agree this could be solved at the search engine / platform level
@jeremycherfas Thanks, very provocative, thoughtful article. One of the links, "fetishization", led me to yet another thoughtful article. Jackpot!
Hasn't the web always been vast though? Adding to the vastness doesn't necessarily affect us. What matters, I imagine, is how tools we rely on like Twitter and Google treat automated material.
another possible direction: we develop models trained to classify identities online as either human or AI as a sophisticated content filter personally I’m drawn to specific human stories exchanged in smaller communities - the public forum may always have been temporary
@jeremycherfas A much needed must read! "Our new challenge as little snowflake humans will be to prove we aren't language models. It's the reverse turing test."
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.