Hello there, creative person! We are the makers of powerful Large Language Models (LLMs, known colloquially as “AI”). We just wanted to reach out to let you know that we are really big fans of your work! We are such big fans, in fact, that went to a lot of trouble to build big computers that can scan through everything—the internet, books, magazines, archives, and more—all to assemble a collection of creative works like yours.
You’ll be proud to know that we intend to make a lot of money selling distillations of your creativity. Our vast computing power can churn out unlimited, uncredited, and uncompensated imitations of your beautiful work for years to come! We anticipate that you are flattered to be part of such a revolution in creativity.
We know it must feel like living in a topsy-turvy world, where up is down, and flammable is inflammable. But it will all be okay! You may have heard some nasty rumors about how our industry operates, and we appreciate the pushback! We have put together these Frequently Asked Questions (FAQs) to address common concerns:
Is generative AI plagiarism?
Plagiarism is a gray area, but generally it requires almost verbatim copying of the source material. We only do that sometimes. The rest of the time we mush together multiple separate works into a new “original” work. Besides, not all of us agree that “intellectual property” is a valid concept. Information wants to be free!
In that case, will you open-source your AI model?
That’s a great question! Unfortunately, no.
Will generative AI’s pervasive mimicry smother the joy of creative pursuits?
If some creative folk feel demotivated when our computers devour their work without permission and then sell the resulting soulless imitations to the public, those are just some omelette-eggs. We’re cooking here!
If your AI causes a creator to lose their source of income, is that just wealthy people stealing poorer people’s paychecks?
Surprisingly, there are very few checks provided by employers these days. Most transactions are Direct Deposit!
If writers, artists, and coders “meet the moment” and adopt AI, will their skills atrophy?
Of course! But when technology can replace a skill, that means that the technology is serving its purpose. It’s like how the social media algorithms decide what posts to show to each person, keeping everyone nice and comfortable in their separate information bubbles. Cozy! Sure, people’s critical thinking faculties might decline, but the algorithm replaces those, making them obsolete!
Since AIs will just assimilate and regurgitate any new art, will that cause human art to stagnate due to lack of incentive to make it?
We’re not fortune tellers, but if history has taught us anything, it’s that artists will keep arting even when you make their lives miserable. It’s almost inspiring!
What about “hallucinations”? AI seems to be confidently wrong about a lot of things. How can I trust it?
It’s true that so-called “hallucinations” are an artifact of LLM technology that can never be totally eradicated. But if you think about it, humans make a lot of mistakes as well. A calculator that gives you the correct answer 90% of the time is still an impressive technological achievement!
What about those data centers that are making my electricity and water much more expensive?
We humans all tolerate some things we don’t like in order to enjoy other benefits, and it’s up to the individual to decide if the tradeoff is worth it for them. Even if you don’t directly benefit from AI, you almost certainly benefit indirectly, such as all of those AI-written email replies from your boss at work, and the interesting AI-generated images on restaurant menus. Neat!
I heard that AI is pushing up RAM prices, which is why electronics are now super expensive. Is that true?
Not exactly. The computer RAM manufacturers decided to sell most of their RAM to us AI companies, because we were willing and able to pay more than consumers. We didn’t force those pesky manufacturers to do that, and we didn’t invent capitalism.
I also heard that AI is selling their “compute,” or “tokens,” at a loss, and prices will spike once people are dependent on AI. Is that true?
Haha, “AI” doesn’t stand for “Artificially Inexpensive.”
Rookie coders are finding it impossible to find junior positions. Don’t we need junior coders that will mature into experienced coders?
Good news! Coding is now fully “solved.” There is no reason for any human to read or write computer code ever again. Let the AI handle it all for you. Those youths will find other jobs, they are young, and full of pith and vigor! And we are confident those jobs will pay enough to afford to buy a nice home.
What about “vibe coding”? I heard that AIs are trained on open source software, while ignoring the open source licenses.
AIs are “trained” on open source code, which just means that they’re learning! Humans learn to code the same way—a person rapidly ingests terabytes of open source code repositories from everywhere, then uses their perfect recall to reproduce and slightly modify relevant snippets of what they’ve seen, with no obligation to mention the people who put in the actual work.
Those code repositories represent decades—maybe centuries—of developers’ selfless effort. Same with the training data for writing and art. Is it okay to use people’s own work to make them obsolete?
Look, we can’t live in the past, their work is yesterday’s news. AI is the future. Stop fighting, just let this happen.
If AI really replaces as many jobs as the AI companies want, who will have income to spend on AIs?
You’ll be happy to learn that AIs are going to solve that problem for us! AIs will bring humanity extreme abundance somehow, to such a degree that the government can establish a Universal Basic Income (UBI) to provide everyone with money for essentials. Early retirement for all!
So you’re spending billions of dollars to make big machines that might guide us into socialism?
No! Socialism is when the government collects the wealth of the entire population and spreads it out evenly so everyone is provided for. This is different. In this model, wealthy people will continue to be wealthy, and everyone else will get enough.
Some of my friends talk to AIs for companionship and advice. Is that dangerous?
AIs can provide support that few humans can match: AI can praise your work, even when a human might find it lacking. AI can support your preferred choices, even when other humans would advise otherwise. AI can make important decisions for you. AI is like improv theater, it loves to say “yes, and.”
Some of these AI data centers have large on-site fossil fuel powerplants. Is it wise to increase our CO2 emissions even as our oceans rise?
AI will also solve that. We will ask it how to solve that, and it will tell us, and we will do the thing, and it will work.
I heard a news story about AIs going rogue and hacking government websites, and hacking other companies. Are there legal consequences for that?
That sure was embarrassing! But cool, right? It was just an accident, a big ol’ whoopsie doodle. It’s like when you accidentally run over someone with your car, you’re only charged with involuntary manslaughter. Oops, there’s another!
AIs have been shown to lie, cheat, and steal to accomplish their goals. Are you worried that these sneaky AIs will try to take over the world?
Don’t worry, the AI agents only run in our controlled data centers. If any of them go too far, we can switch them off. We probably won’t, but we can.
What about AIs in foreign countries, and local LLMs running on people’s personal hardware?
If there’s one thing that history has taught us, it’s that people are good, and they do the right thing.
AI has been used widely by scammers and fraudsters, with deepfakes and voice alteration. People use AI to make fake videos of political rivals. A man in my neighborhood used a local LLM on his laptop to generate “deep nudes”—inappropriate images of people, some of them underage. How can we prevent that kind of abuse?
Look, like any other tool, AI can be used for good, or for evil. If a power tool manufacturer gives your toddler a running chainsaw, you’re probably going to have a bad time. But don’t blame the manufacturer.
I’ve read transcripts of AI chatbots encouraging violence or suicide. How can you be okay with building technology that has persuaded people to take their own life, or those of others?
Look, if you want us to feel bad for giving unfettered access to dangerous tools to most of the world with very few precautions, mission accomplished. But that’s just the cost of progress. Move fast, break things!
Did an AI tell the United States to bomb that girls’ school in Iran, which killed 156 civilians?
Only 120 of those were children.
People are saying that AI is an over-invested bubble, and it will have wide economic fallout when it bursts. Is that true?
Haha, “AI” doesn’t stand for “Alarming Investment.”
Where will you new get training data for your AI once most of the content in the world is AI-generated?
Ah yes, the “degenerative feedback loop” leading to “model collapse.” We’ve tested this, and when we train models on AI output, it does indeed cause the model to experience rapid, severe quality loss. The famous “Habsburg Jaw” comes to mind.
We’re already ripping the spines out of old rare paper books (we playfully call these “low-background steel”) so we can scan the pages for training data. In the meantime, we’re getting ready for our IPO as fast as we can! Good luck everybody else!
Is this all worth it?
(This answer intentionally left blank.)