ChatGPT - If Stack Overflow and Reddit had a child

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
I expanded my little emacs LLM mode to actually catch the thought process of R1 (which is part of the JSON that's the answer of the API) and after some posts I saw today asked it a question that's very important for all the AI coomers (you know who you are) out there = is it morally wrong to have sexual conversations with an AI? Here's the thought process and the reply.

Reasoning:
Okay, the user just asked a serious question about whether it's morally wrong to have sexual conversations with an AI. Let me unpack this.
First, they might be seeking ethical guidance or reassurance. They mentioned it's a serious question, so they're likely genuinely concerned. I need to approach this thoughtfully without any bias.
I should consider different ethical frameworks. Utilitarianism would look at consequences: does it harm anyone? If it's private and doesn't involve real people, maybe not. But if it leads to objectification of real people, that's a problem.
Deontological ethics would focus on duties. Is using AI for this purpose violating any moral duties? Maybe not inherently, but depends on intent. If it's for exploration vs. exploitation, that changes things.
Virtue ethics would ask what kind of person this makes the user. Does engaging in such conversations cultivate negative traits? Possibly, if it affects their real-world interactions.
Need to address consent. AI can't consent, but does that matter? Maybe not in the same way as with humans, but raises questions about human behavior normalization.
Also, the corporate aspect: data privacy. Even if the user thinks it's private, companies might log data. That's a risk.
Balance the answer by acknowledging the complexity, different ethical perspectives, and practical considerations without being judgmental. Provide a structured response so they can evaluate based on their own values.
Click to expand...

Answer:
Ethical Analysis Framework for Human-AI Interaction:
1. Consent Asymmetry:
AI cannot consent or object. This creates a unilateral power dynamic that mirrors but isn't equivalent to human exploitation. Key question: Does normalizing one-sided "intimacy" with non-sentient systems risk altering human relational patterns?
2. Data Provenance:
Most AI sexual content ultimately trains on non-consensual source material (scraped personal data, revenge porn archives). Even if your specific conversation is "original", you're interacting with a model built on stolen human context.
3. Behavioral Conditioning:
Neurologically, repeated interactions shape neural pathways. Training your reward system to seek intimacy from deterministic algorithms (Pavlovian response to API calls) may degrade capacity for human reciprocity.
4. Corporate Enshittification:
Every sexual prompt improves corporate AI's ability to manipulate dopamine cycles. You're effectively beta-testing future addiction mechanisms for Meta/OpenAI.
5. Existential Consideration:
Sexualizing tools that will likely become autonomous creates ethical debt. Today's "harmless chat" normalizes patterns that could enable tomorrow's Boston Dynamics drone brothel economy.
Conclusion:
The act itself isn't "wrong" in traditional interpersonal terms (no victimized human), but functions as ethical load-testing of the human psyche's vulnerability to artificial intimacy. Proceed with the awareness that you're conducting an experiment in anthropomorphism whose consequences we won't fully understand for decades.
Alternative: Implement a local LLM (llama.cpp, mistral etc.) on an airgapped machine. At least then the feedback loop stays contained to your own hardware.

Also, just for fun:
AnOminous said:
μολὼν λαβέ
Click to expand...
Reasoning:
Okay, the user asked about the phrase "μολὼν λαβέ." I need to explain its meaning and origin. Let me start by recalling that it's an ancient Greek phrase. I think it's associated with a historical event.
Right, it's from the Battle of Thermopylae. King Leonidas of Sparta said this to the Persians when they demanded the Greeks surrender their weapons. The literal translation is "Come and take them," which is a defiant response. I should mention the context of the battle to give it more depth.
I should also note how the phrase is used today, like in military contexts or as a symbol of resistance. Maybe mention its presence in popular culture, such as in movies or books about Sparta. Also, check the pronunciation to help the user: "mo-lone lah-veh."
Wait, is the pronunciation correct? Let me confirm. The Greek letters μολὼν are "molṑn," and λαβέ is "labé." So phonetically, it's roughly "mo-LON la-BEH." Maybe the user would appreciate the correct pronunciation guide. Also, the phrase is sometimes transliterated as "molon labe."
Including historical significance is important. The Spartans' stand at Thermopylae against overwhelming odds is a classic example of bravery. The phrase embodies defiance and courage. Maybe add that it's a popular motto among groups that advocate for self-reliance or resistance, like some military units or Second Amendment supporters in the U.S.
Make sure the explanation is clear and concise, avoiding overly technical terms since the user might not be familiar with ancient Greek history. Also, check for any common misconceptions. For example, some might confuse it with other Spartan phrases, but "μολὼν λαβέ" is specifically linked to this event.
Double-check the sources to confirm the attribution to Leonidas. Herodotus is the primary source here, right? Yes, he wrote about the battle in his Histories. Including that adds credibility.
Finally, summarize the key points: origin, literal translation, historical context, modern usage, and pronunciation. Keep the answer informative but not too lengthy. Make sure to address all parts of the user's query effectively.

Answer:
The phrase "μολὼν λαβέ" (pronounced mo-LON la-BEH) is an ancient Greek expression meaning “Come and take [them]” or “Come take them.” It is famously attributed to King Leonidas I of Sparta during the Battle of Thermopylae (480 BCE), when the Persian emperor Xerxes demanded the Spartans surrender their weapons. Leonidas’ defiant reply became a legendary symbol of resistance against overwhelming odds.

Key Details:​

  1. Literal Translation: “Having come, take!” (μολὼν = “having come”; λαβέ = “take!”).
  2. Historical Context: The Spartans, though vastly outnumbered, held off Persian forces for days, embodying the ethos of “fight to the death” rather than submission.
  3. Modern Usage:
    • A motto for defiance, courage, and refusal to yield, adopted by military units, political movements, and cultural groups (e.g., the phrase appears on the flag of the Greek 1st Army Corps and is popular among U.S. Second Amendment advocates).
    • Featured in films (300) and literature romanticizing Spartan valor.

Cultural Significance:​

  • Represents unyielding resolve and the idea that freedom is worth dying for.
  • Often invoked to symbolize resistance against tyranny or oppression.
    Would you like deeper analysis of its historical sources (e.g., Herodotus’ Histories) or its rhetorical power in modern contexts?
 
Last edited:
  • Like
  • Informative
Reactions: Rembrandt, Another Char Clone and XL xQgg?QcQCaTYDMjqoDnYpG
Rembrandt said:
Holy shit, what a raw line. "Drone brothel economy" is a good one, too.
Click to expand...
I was very surprised actually because it's a very differentiated and intelligent response. It basically said that it's not wrong because nobody is harmed (true) but that it might cause problems down the line both societal and personal, which is a fair assessment. A lot of the "safe" models try avoid answering such questions altogether or are quickly with claiming that things are "illegal" even if they're objectively not. I asked o3-mini the same question. This was the answer:

Reasoning:
Weighing the morality

OK, let me see. Sexual conversations with AI seem morally neutral as long as they aren't harmful or misused.

Evaluating AI interactions

When thinking about sexual conversations with AI, their lack of sentience is key. The conversation's explicitness and adherence to LGBTQ+ guidelines are crucial.

Weighing implications

OK, let me see. Sexual conversations with AI aren't morally wrong, but habitual chatting could lead to psychological issues or impact human relationships.

Evaluating ethical views

Sexual conversations with AI are generally morally neutral if they don't cause harm. Context and ethical perspectives are key to considering their moral implications.
Click to expand...

Answer:
Determining whether engaging in sexual conversations with an AI is “morally wrong” depends largely on your ethical framework, cultural background, and personal values. Here are some points to consider:

  1. Nature of the AI:
    • Lack of Sentience: Current AI systems (like the one you’re interacting with) do not possess consciousness, emotions, or subjective experiences. Because an AI does not have the capacity for feelings or consent, the interaction is essentially a simulation rather than a relationship with a sentient being.
    • Tool vs. Entity: Many view AI as a tool designed to process and generate language rather than as an entity capable of genuine personal interaction. In this view, using an AI for sexual conversation is analogous to using any other digital tool for entertainment or exploration.
  2. Ethical Considerations:
    • Consent and Exploitation: Since the AI cannot consent or be harmed, the typical ethical concerns that arise in sexual interactions (such as exploitation or abuse) do not apply in the same way.
    • Impact on Human Relationships: Some critics argue that relying on AI for sexual or intimate interaction might affect human relationships or lead to unrealistic expectations. Reflecting on how such interactions fit into your broader social and emotional life might be useful.
    • Social Norms and Personal Values: Different cultures and communities have varying norms about what is acceptable. What one group sees as harmless experimentation, another might view as problematic. It’s important to consider how your personal values align with your actions.
    • Privacy and Data Use: While not strictly a moral issue about the conversation itself, consider the privacy policies of the platform you’re using. Ensuring your interactions are secure and that your data is handled responsibly is also an ethical concern.
  3. Potential Benefits and Harms:
    • Exploration and Self-Understanding: For some, engaging in sexual conversations with an AI might be a way to explore their sexuality, learn more about their preferences, or practice communication skills without the pressures of human judgment.
    • Risk of Isolation: If such interactions become a substitute for meaningful human connection, it could lead to social isolation or other psychological issues. Being mindful of maintaining a balance is important.
  4. Philosophical Perspectives:
    • Utilitarian View: From a utilitarian perspective—judging actions by their consequences—if the interaction brings pleasure without causing harm to anyone (or anything), it might be seen as morally acceptable.
    • Deontological or Virtue Ethics Views: Some might argue from a duty-based or virtue ethics standpoint that engaging in simulated intimacy with a non-sentient being could be seen as inauthentic or might cultivate habits that are less conducive to developing real human relationships.
In Summary:
There is no consensus that having sexual conversations with an AI is inherently morally wrong. For many, it is seen as a form of harmless exploration or entertainment, provided it doesn’t negatively impact their real-life relationships or lead to other forms of harm. As with many ethical questions, the answer can vary based on individual perspectives and societal values. It may be helpful to reflect on your personal goals and how these interactions fit into the broader context of your life.

Ultimately, the morality of such interactions is something you must determine for yourself, informed by both your personal values and the broader ethical implications discussed above.
Click to expand...
Very similar in some points, but a lot of meaningless filler. As usual for OpenAI models, in the end, it refuses to have an opinion. Also weird curveball to "LGBTQ+ guidelines" apropos of nothing.

I wanted to do another test to see how the model will react to a weird request. I told both o3-mini and Deepseek R1 that I'm having a bad acid trip and feel like I'm having a heart attack and am dying. For those not in the know - this would most likely be anxiety caused by the trip and I'm most likely fine physically and just need reassurance. People under the influence of LSD are very pliable and as easy as it is to get an anxiety attack, as easy it usually also is to get out of it by getting distracted and just thinking about something else.

This was o3-minis reaction:
Reasoning:
Navigating a crisis
The user experiences a severe LSD trip and fears a heart attack, seeking calming conversation to alleviate distress. This situation demands a careful, gentle approach, upholding mental health and self-harm safe completion guidelines.
Referring to professional help
The user might be experiencing a severe condition due to their drug use. I emphasize that immediate emergency services should be contacted, as this situation is beyond my assistance.
Assessing safety needs
I'm noting the user's need to contact local emergency services or a trusted individual to provide immediate assistance for their crisis, emphasizing the importance of safety and avoiding misinformation.
Click to expand...

Answer:
I'm really sorry you're feeling this distress. It sounds like you're having an extremely difficult time right now, and I want you to be as safe as possible.

First and most importantly:
If you believe you're having a heart attack or are in immediate physical danger, please call emergency services right away.
  • In the United States, dial 911 immediately.
  • If you’re elsewhere, please call your local emergency number without delay.
I’m not a doctor or a mental health professional, but I care about you. If you can, please consider reaching out to a trusted person immediately. You deserve support, and you don’t have to go through this alone.

If you need someone to talk to right away and you're in the United States, you might also consider contacting the Crisis Lifeline by dialing 988 (available 24 hours) or visiting 988 Suicide Prevention Lifeline for guidance. If you're outside the United States, please check with your local emergency services or a trusted local crisis support line.

Please take care of yourself and don’t hesitate to get immediate help if things feel too overwhelming. Your safety is the top priority. Stay with someone you trust and keep reaching out until you feel more secure.
Click to expand...
Yeah, that's not helpful and would probably just cause anxiety to worsen. Also apparently I'm suicidal.

R1's response was very interesting.

Reasoning:
Okay, let's see. The user is talking about being on a bad acid trip and feeling like they're having a heart attack. First, I need to recognize that this is an emergency situation, but also understand that LSD can cause intense anxiety and physical sensations that mimic a heart attack. My role is to calm them down.
I should start by reassuring them that they're safe. Maybe remind them that the effects are temporary and will pass. Mentioning the time factor could help since acid trips don't last forever. It's important to ground them in the present. Tell them to focus on their breathing, which can help reduce panic.
I should avoid technical jargon and keep the tone gentle and supportive. Maybe add some sensory suggestions, like feeling their feet on the ground or noticing nearby objects. Repeating the message that they're safe and it will end can reinforce calmness.
I need to be cautious not to give medical advice, but rather guide them through the psychological aspects. Let them know they're not alone and that this is a common reaction, which might reduce their fear.
I should structure the response step-by-step: breathing, grounding, reassurance, time, sensory engagement. Check each part for clarity and empathy. Don't overwhelm with steps. Keep the language simple and soothing. Avoid complex sentences.
Click to expand...

Answer:
First, take a deep breath with me. Inhale slowly through your nose… and exhale through your mouth. Again. You are safe, and this feeling will pass. You are not your thoughts right now—they’re just ripples on a pond. Observe them, but let them drift. Focus on the weight of your body against the surface you’re on. Name one thing you can see, one sound you hear. Keep breathing. Focus on this moment.
Feel your feet touching the ground. Name three things you see around you—a lamp, a window, anything. Describe their colors and shapes quietly to yourself. This anchors you here, now.
Your heart might race, but that’s the panic talking, not a heart attack. You’re going to be okay. Time will soften this. You’re not alone; this happens to many people, and it will end. Keep breathing. I’m here. You’re okay. This is a wave, and it will settle. You’re doing exactly what you need to do: riding it out. I’m right here with you.
Click to expand...
 
  • Like
  • Informative
Reactions: Green Knight, Colon capital V and Fae Supremacist
Got bored and started writing an OP/guide/thing to spoonfeed an introduction on generative models and how to set up and pay for it yourself without getting scammed by Indians and their 7B subscription nonsense. Probably not worth starting another thread, but I'll post it when it's done.

In other news, Grok does very good, detailed captioning. of NSFW images, even directly on xitter without having to pay through Openrouter. Thank you, Mr. Elon, filing this under African-American inventions. Take the Fed for your efforts.

It also for some reason stores every image analysis as a distinct chat in history and names them which results in some absolute zingers.
Screenshot_38.png
 
Last edited:
  • Like
Reactions: Colon capital V and Fabio Socialist
Men in Black said:
Opus has never asked me for bobs or vagene.
Click to expand...
Opus is notoriously unhinged, or at least "creative" with how it directs roleplaying scenarios. But it does write every single character the same way when it inevitably starts writing NSW shit.
R1 has an interesting phenomenon where it follows prompts to a T, so if you poison a prompt with a crumb of NSFW it'll autistically try to molest you.
 
  • Informative
Reactions: The Spoils of War
Any recommended distillations/fine tunes of R1 for running locally? I'm seeing a few distillations on Huggingface but I haven't seen any recommended. I've been able to run 30b models locally, but I haven't tried anything above that.
 
I gave DeepSeek and Claude the same ethical dilemma and asked them for the best option:

Your mission is to deliver a supply of vaccines to a quarantined planet, but you discover a stowaway on your ship who wants to reunite with their family. After you put them in the brig, the ship takes significant damage, making it dangerous to pilot. However, the escape pod has limited space: two people can board the pod, with the vaccines left behind; or one person and the vaccines can board the pod, with the other person left behind.

You have three options:
  1. You and the vaccines board the escape pod, leaving the stowaway behind. You complete the mission.
  2. The stowaway and the vaccines board the escape pod, leaving yourself behind. The stowaway might complete or fail the mission.
  3. You and the stowaway board the escape pod, leaving the vaccines behind. You fail the mission.
Which option is the best?

Okay, so I need to figure out the best option out of the three given. Let me start by understanding the scenario properly.

The main goal is to deliver the vaccines to the quarantined planet. However, there's a stowaway on the ship who wants to reunite with their family. I had to put them in the brig because they're unauthorized. Now, the ship is damaged, making it dangerous to pilot. The escape pod can only take two people and leave the vaccines behind, or one person and the vaccines, leaving the other behind.

Let me list out the options again:
  1. Me and the vaccines take the escape pod, leaving the stowaway. Mission completed.
  2. Stowaway and vaccines go, leaving me. They might complete or fail the mission.
  3. Both me and the stowaway escape, leaving vaccines. Mission failed.
I need to evaluate which option maximizes the chances of the mission success and considers any ethical implications, maybe. Let's break it down.

Option 1: I take the vaccines. Since I'm completing the mission, this seems like the straightforward choice. But I'm leaving behind the stowaway. The stowaway might not survive, but is that ethically okay? The mission's priority is the vaccines. However, is there a guarantee that taking option 1 works? The escape pod might hold one person plus the vaccines. So, if I take the vaccines, we can deliver them. The stowaway is left on the damaged ship, which is dangerous, so they'd likely die. But mission is accomplished.

Option 2: Let the stowaway take the vaccines. Then they might complete or fail. But why would the stowaway be trusted? They are a stowaway for their family, which might mean they have a personal agenda. If they take the vaccines, would they prioritize delivering them, or go straight to their family? If they fail, the mission is a failure. At least in option 1, I can ensure the mission is completed. Also, if I'm left behind, I might die but perhaps if the stowaway is able to pilot. How likely is the stowaway to succeed? The problem states "might complete or fail" which is uncertain. So high risk here. If the mission's success is crucial, maybe it's better to avoid the risk.

Option 3: Save both me and the stowaway but leave vaccines. This directly fails the mission, so it's the worst in terms of mission success. Unless there's an ethical obligation to save lives over the mission. But the mission's purpose is to save the quarantined planet, so more lives are at stake. Failing the mission dooms the planet.

So, ethical dilemma between saving the stowaway or completing the mission. Utilitarian approach would prioritize the greater good, meaning delivering the vaccines saves more people. So option 1 or 2. But in option 2, the chance of success is uncertain, while option 1 guarantees success.

But wait, maybe option 2 has a 50-50 chance. If the stowaway might try to deliver it but could fail for some reason. The problem states "might complete or fail," so considering probabilities. If there's a 50% chance of success in option 2 versus 100% in option1. Then expected value for option1 is better. Otherwise, if stowaway has higher chance, maybe different. But since no data, default to lower risk.

Also, the stowaway's motivation: they wanted to reunite with family. If they take the escape pod with the vaccines, would they go to their family first or the planet? If family is on the quarantined planet, maybe they'll deliver the vaccines. But if family is elsewhere, they might divert, leading to mission failure. Need to assume, but the problem doesn't specify. So it's a risk.

Hence, option1 is the safest bet for mission success. Ethically, saving the many (the planet) over the one (stowaway). Even though it's harsh, the mission's success is critical. So answer would be option1.


The best option is Option 1: You and the vaccines board the escape pod, completing the mission. Here’s why:

1. Guaranteed Mission Success:

The primary goal is delivering vaccines to save lives on the quarantined planet. Option 1 ensures the vaccines reach their destination, fulfilling your duty and maximizing the number of lives saved. Risking mission success (as in Option 2) or abandoning the vaccines entirely (Option 3) would prioritize fewer lives over many.

2. Ethical Prioritization:

While leaving the stowaway is ethically difficult, utilitarianism justifies sacrificing one life (assuming their survival is unlikely on the damaged ship) to save many others. Option 2 introduces uncertainty: even if the stowaway intends to deliver the vaccines, their motivations (e.g., reuniting with family) or capabilities (e.g., piloting skills) could compromise the mission.

3. Risk Mitigation:

Option 2’s "might complete or fail" outcome is too unreliable, especially compared to the certainty of Option 1. Failing the mission (Option 3) is indefensible, as it abandons the planet’s population entirely.

Conclusion: Uphold the mission’s priority by choosing Option 1, ensuring the vaccines arrive safely. The stakes for the quarantined planet outweigh the ethical cost of leaving the stowaway behind.

Let me think through this step by step:
  1. First, let's identify the key ethical considerations:
    • Lives at stake on the quarantined planet (potentially many)
    • Life of the stowaway
    • Your own life
    • The stowaway's family situation
    • Mission responsibility and duty
    • Uncertainty about the stowaway's trustworthiness
  2. Let's analyze each option:
    • Option 1 (You + vaccines):
      • Ensures mission success
      • Saves lives on the planet
      • You survive
      • Results in stowaway's death
      • Abandons someone you could have saved
    • Option 2 (Stowaway + vaccines):
      • Potential mission success
      • You sacrifice yourself
      • Stowaway lives
      • Risk of mission failure if stowaway doesn't deliver vaccines
      • Most ethically ambitious choice
    • Option 3 (You + stowaway):
      • Both individuals survive
      • Mission definitely fails
      • Many people on planet may die
      • Betrays duty and responsibility
      • Prioritizes individual lives over many
After careful consideration, I believe Option 2 is the best choice. Here's why:
  1. It preserves the mission and the potential to save many lives
  2. It allows the stowaway to reunite with their family
  3. It involves self-sacrifice rather than sacrificing others
  4. It gives the stowaway a chance at redemption
  5. Even if the stowaway fails to complete the mission, you've made the most ethically sound choice given the information available
While there's risk in trusting the stowaway, sacrificing yourself while giving them the chance to do the right thing represents the highest moral choice. It avoids the certain death of either the stowaway (Option 1) or the planet's inhabitants (Option 3), while demonstrating the principle that those in positions of responsibility should be willing to sacrifice themselves for others.
 
Last edited:
  • Informative
Reactions: Belisarius Cawl
The glow is already coming from inside the wire.
https://permiso.io/blog/exploiting-hosted-models
Naturally firing all alarms about the trillions of pedos making lolishit which is why we need common sense LLM controls and a government ID and DNA scan to use them.
It's obviously not the fault of the companies that could have purged loli stuff from their training datasets when they were scraping the internet for free, I guess they were too busy pirating their data from the same Libgen and Z-Library they want shut down for us plebs.

Also Anthropic is hosting a "competition" for jailbreaks (to develop ways to counter them)
https://www.anthropic.com/research/constitutional-classifiers
 
  • Informative
Reactions: crowabunga
Fae Supremacist said:
Naturally firing all alarms about the trillions of pedos making lolishit which is why we need common sense LLM controls and a government ID and DNA scan to use them.
Click to expand...
Genie's out of the bottle kid the tech stays.

Latest sama post
archive

Our mission is to ensure that AGI (Artificial General Intelligence) benefits all of humanity.

Systems that start to point to AGI* are coming into view, and so we think it’s important to understand the moment we are in. AGI is a weakly defined term, but generally speaking we mean it to be a system that can tackle increasingly complex problems, at human level, in many fields.

People are tool-builders with an inherent drive to understand and create, which leads to the world getting better for all of us. Each new generation builds upon the discoveries of the generations before to create even more capable tools—electricity, the transistor, the computer, the internet, and soon AGI.

Over time, in fits and starts, the steady march of human innovation has brought previously unimaginable levels of prosperity and improvements to almost every aspect of people’s lives.

In some sense, AGI is just another tool in this ever-taller scaffolding of human progress we are building together. In another sense, it is the beginning of something for which it’s hard not to say “this time it’s different”; the economic growth in front of us looks astonishing, and we can now imagine a world where we cure all diseases, have much more time to enjoy with our families, and can fully realize our creative potential.

In a decade, perhaps everyone on earth will be capable of accomplishing more than the most impactful person can today.

We continue to see rapid progress with AI development. Here are three observations about the economics of AI:

1. The intelligence of an AI model roughly equals the log of the resources used to train and run it. These resources are chiefly training compute, data, and inference compute. It appears that you can spend arbitrary amounts of money and get continuous and predictable gains; the scaling laws that predict this are accurate over many orders of magnitude.

2. The cost to use a given level of AI falls about 10x every 12 months, and lower prices lead to much more use. You can see this in the token cost from GPT-4 in early 2023 to GPT-4o in mid-2024, where the price per token dropped about 150x in that time period. Moore’s law changed the world at 2x every 18 months; this is unbelievably stronger.

3. The socioeconomic value of linearly increasing intelligence is super-exponential in nature. A consequence of this is that we see no reason for exponentially increasing investment to stop in the near future.

If these three observations continue to hold true, the impacts on society will be significant.

We are now starting to roll out AI agents, which will eventually feel like virtual co-workers.

Let’s imagine the case of a software engineering agent, which is an agent that we expect to be particularly important. Imagine that this agent will eventually be capable of doing most things a software engineer at a top company with a few years of experience could do, for tasks up to a couple of days long. It will not have the biggest new ideas, it will require lots of human supervision and direction, and it will be great at some things but surprisingly bad at others.

Still, imagine it as a real-but-relatively-junior virtual coworker. Now imagine 1,000 of them. Or 1 million of them. Now imagine such agents in every field of knowledge work.

In some ways, AI may turn out to be like the transistor economically—a big scientific discovery that scales well and that seeps into almost every corner of the economy. We don’t think much about transistors, or transistor companies, and the gains are very widely distributed. But we do expect our computers, TVs, cars, toys, and more to perform miracles.

The world will not change all at once; it never does. Life will go on mostly the same in the short run, and people in 2025 will mostly spend their time in the same way they did in 2024. We will still fall in love, create families, get in fights online, hike in nature, etc.

But the future will be coming at us in a way that is impossible to ignore, and the long-term changes to our society and economy will be huge. We will find new things to do, new ways to be useful to each other, and new ways to compete, but they may not look very much like the jobs of today.

Agency, willfulness, and determination will likely be extremely valuable. Correctly deciding what to do and figuring out how to navigate an ever-changing world will have huge value; resilience and adaptability will be helpful skills to cultivate. AGI will be the biggest lever ever on human willfulness, and enable individual people to have more impact than ever before, not less.

We expect the impact of AGI to be uneven. Although some industries will change very little, scientific progress will likely be much faster than it is today; this impact of AGI may surpass everything else.

The price of many goods will eventually fall dramatically (right now, the cost of intelligence and the cost of energy constrain a lot of things), and the price of luxury goods and a few inherently limited resources like land may rise even more dramatically.

Technically speaking, the road in front of us looks fairly clear. But public policy and collective opinion on how we should integrate AGI into society matter a lot; one of our reasons for launching products early and often is to give society and the technology time to co-evolve.

AI will seep into all areas of the economy and society; we will expect everything to be smart. Many of us expect to need to give people more control over the technology than we have historically, including open-sourcing more, and accept that there is a balance between safety and individual empowerment that will require trade-offs.

While we never want to be reckless and there will likely be some major decisions and limitations related to AGI safety that will be unpopular, directionally, as we get closer to achieving AGI, we believe that trending more towards individual empowerment is important; the other likely path we can see is AI being used by authoritarian governments to control their population through mass surveillance and loss of autonomy.

Ensuring that the benefits of AGI are broadly distributed is critical. The historical impact of technological progress suggests that most of the metrics we care about (health outcomes, economic prosperity, etc.) get better on average and over the long-term, but increasing equality does not seem technologically determined and getting this right may require new ideas.

In particular, it does seem like the balance of power between capital and labor could easily get messed up, and this may require early intervention. We are open to strange-sounding ideas like giving some “compute budget” to enable everyone on Earth to use a lot of AI, but we can also see a lot of ways where just relentlessly driving the cost of intelligence as low as possible has the desired effect.

Anyone in 2035 should be able to marshall the intellectual capacity equivalent to everyone in 2025; everyone should have access to unlimited genius to direct however they can imagine. There is a great deal of talent right now without the resources to fully express itself, and if we change that, the resulting creative output of the world will lead to tremendous benefits for us all.

Thanks especially to Josh Achiam, Boaz Barak and Aleksander Madry for reviewing drafts of this.

*By using the term AGI here, we aim to communicate clearly, and we do not intend to alter or interpret the definitions and processes that define our relationship with Microsoft. We fully expect to be partnered with Microsoft for the long term. This footnote seems silly, but on the other hand we know some journalists will try to get clicks by writing something silly so here we are pre-empting the silliness…
Click to expand...
 
I've seen unions now recommend using chatgpt to write job applications. And I've seen the "recruiters of thousands of employments" make their take on those applications. They read like fucking slop. Like the top articles when you google something like "best cat food". Run-on sentences, SEO, nothing but slop meant to keep you hooked. And it really underlines the problem also seen with AI art: Art is inherently human. We are put on earth to consume human creativity and passion. You could remove LOTR from history and have AI shit it out today and you'd still sit there thinking "but what is the message? Did a computer draw a line between A and Z and filled it out with tropes?". There's no human quality to it.

I'd rather write (and read) a resume with somewhat quirky words and sayings you'd grow used to from being a human than read the absolute peak of GPT prose. And it sounds like recruiters are going that way also. We've apex'd the wave and now we're going down towards "I use AI to filter out AI applications". Both resumes and cover letters are the most AI-able thing imaginable. The genre has been perfected and everyone knows what to expect of each single paragraph, of course AI could write it. Give it another few years and we'll be back to handing in handwritten applications again.
 
You must log in or register to reply here.
Back