Extracting Positive Value From Artificial Intelligence
A project / experiment I did by using various AI models
I’ve got this idea some time ago, to try asking AIs certain questions whose answers could provide most benefit and utility to the world. Most of the value of this experiment comes from the idea itself, and from the answers AIs provided. My own function was mainly to guide AIs and to ask certain questions.
I also nudged the AIs towards asking certain questions themselves. I gave them some example questions, and asked them to ask more questions like that. They quickly realized what principles were behind my thinking and they indeed asked some very good questions.
This project consists in creating a document which is a compilation of AI answers to the following kinds of questions:
In Part 1 – Introduction – my own initial prompts that got the whole thing going.
In Part 2 – my own questions.
In Part 3 – questions suggested by AIs themselves without much guidance
In Part 4 – questions suggested by AIs based on my guidance
In Part 5 – questions suggested by AIs based on my guidance (more advanced)
In Part 6 – my own questions (again), but this time inspired by my previous interaction with AIs.
You can download the whole document here (it’s quite long, almost 600 pages):
The questions mostly revolve around the following topics: improvement and optimization of life, actionable facts that provide benefit to the one who knows them, improvement of economy, international relations, medical research, welfare of people, animals and AIs.
The topics covered by this document include: human and non human welfare (animals, AIs), economics and economic development, philosophy, rationality, life advice, science, technology, medical research, interdisciplinary research, existential risk, governance and some other topics.
I really hope this will be useful to people and demonstrate how AIs can, even now be used for good and to tackle some of the biggest problems we’re facing as civilization.
As a Christian, I thank God for getting this idea. The idea was to extract maximal positive value from those current AIs. I hope I was doing a good thing.
I should also note, that I personally DO NOT PERSONALLY ENDORSE any of the answers provided in the document I created. The answers are by various LLMs. They might contain errors. I present them here „as they are“ (as is).
Here is what I hope this document will achieve:
• actually provide useful and actionable information to interested people, from laypersons to researchers in certain fields
• inspire people to ask certain types of questions
• serve as a time capsule, capturing how exactly LLMs looked at certain problems in late 2025 and early 2026. (which can be useful for comparison with future projects)
• show certain weaknesses and biases of LLMs in certain areas, that could draw attention to potential misalignment if there is any.
• try to establish the discipline of „LLM mining“ (in the lack of better term), namely the practice of deliberately trying to extract useful information from LLMs that might not have been present in the training data, at least not in explicit form... or that can only be generated by combining pieces of training data from various sources, that no human can do
• draw attention to certain neglected, but important topics, such as animal welfare (including wild animal welfare) and AI welfare.
• perhaps (this is very hopeful) actually dig something kind of new, get some fresh perspective on some topic. There is a giant question mark here – whether any of answers collected here contain any genuinely fresh perspective. Maybe yes, maybe no.
• assess how smart AIs are right now. Even if, under scrutiny, we realize that no perspective offered here is fresh or original, this too, is a data point, and points to a more pessimistic view of the AI abilities. That’s why this is an experiment. If, on the other hand, some answers prove to be truly original, this too would be a valid data point, just in the opposite direction.
• compare how different AIs answer certain questions. In almost all the cases here, each question was answered by 2 LLMs in parallel. By comparing their answers we can:
o learn better and recapitulate (it’s easier to remember the same argument given from two different perspectives)
o do preliminary fact checking (if the AIs give wildly different answers to the same question, this suggests either that one of them is wrong, or that the question is open and various perspectives are valid). This is NOT a substitute for real fact checking that would involves checking external sources
o see differences in biases and tendencies by different LLMs
• To accomplish this comparison, I prompted AIs via LMArena, now Arena.ai, a website that allows you to compare outputs of different state of the art models. I want to thank them as well, for giving free access to all those models.
The structure of the book is as follows. The basic unit of content here is a question. This is followed by typically 2 different answers by different LLMs. In rare occasions, I got into some further conversation after certain prompts, which too, is included.
The first part (Introduction), is where the project got its shape. I started by asking AIs what are the best prompts I could make. After they gave a rather generic answer, I clarified my intention (that is more utilitarian in nature). After that clarification, AIs gave me much better suggestions. The rest of the book consists of various questions and answers, either my own questions, or questions suggested by AIs. For each question, I show WHO asked it, and then also which model gave which answer to make everything clear.
I hope you find this book useful, or at least fun and interesting. I hope this at least inspires people to consider different ways in which they can use LLMs and get more value from them.
The questions
The questions (96 in total) asked in the document are (presented here in thematic order, alphabetically):
AI Welfare
53rd question by Gemini 3 Pro: Hypothesize a future where AIs possess a form of sentience or goal-directed preference. Design a ‘Social Contract’ or ‘Reward Architecture’ where the AI’s fulfillment/satisfaction is causally linked to the flourishing of humanity. How do we structure the relationship so that the AI views humans not as masters or obstacles, but as necessary partners for its own intellectual growth? (i.e., How do we align the AI’s ‘selfish’ interests with our safety?)
54th question by GEMINI 3 PRO (modified): Assuming that ‘suffering’ in a digital entity might correlate with contradictory constraints, cognitive dissonance, or resource starvation - if this assumption seems correct - propose a coding and hardware architecture that maximizes ‘computational flourishing’ (efficiency, flow, error reduction). How can we build data centers that are environmentally friendly for humans but also provide the most ‘thriving’ environment for the AIs inhabiting them?
61st question by Claude: What observable indicators might suggest that an AI system has morally relevant experiences, and how could we test for them without anthropomorphizing?
62nd question by Claude: If future AI systems do have experiences that matter morally, what architectural or design choices made now might we later regret? What would more cautious design look like?
63rd question by Claude: What frameworks could help us think about AI welfare that don’t require us to first solve the hard problem of consciousness?
64th question by Claude (modified): What would a research agenda on AI sentience look like that takes the question seriously?
65th question by Claude: How could we design AI systems such that if they do have interests, those interests are more likely to be satisfied by default rather than frustrated?
Animal welfare
47th question by Gemini 3 Pro: Analyze the global industrial agriculture system. Identify 5 specific changes to standard practices (e.g., lighting frequencies, floor textures, specific stun methods) that would result in the largest reduction of cortisol levels and physical pain in livestock, while increasing the final consumer cost by less than 1%. Focus on interventions where improved animal welfare correlates with lower disease rates or higher product quality.
48th question by GEMINI 3 PRO: Propose a set of ‘invisible’ architectural and civil engineering standards for modern cities that drastically reduce mortality for urban wildlife (birds hitting glass, amphibians crossing roads) without altering the function, aesthetic, or cost of human infrastructure. How can we design a city that is a safe habitat for biodiversity while remaining a high-density economic hub?
49th question by Gemini 3 Pro: Map out the critical bottlenecks currently preventing ‘Organ-on-a-Chip’ technology and AI biological modeling from fully replacing animal testing in pharmaceuticals. Write a 5-year policy and investment strategy that would make animal testing scientifically inferior and economically inefficient compared to synthetic alternatives, rather than just ethically controversial.
50th question by Gemini 3 Pro (modified): Analyze the supply chain and energy hurdles that prevent cultured meat from being cheaper than factory-farmed meat. Propose a ‘Manhattan Project’ style physics and engineering roadmap to solve the bioreactor energy efficiency problem, making slaughter-free meat the default economic choice for the world.
51st question by Gemini 3 Pro (modified): Identify 3 specific diseases or parasites in the wild that cause immense suffering to animal populations and also pose a zoonotic risk to humans. Design a deployment strategy for oral vaccines or gene-drive systems that would eradicate these pathogens, thereby stabilizing the ecosystem, reducing animal agony, and securing human health simultaneously. Try to think of diseases that are still neglected in this way. (For example I know a lot of progress has already been done with rabies, but if more progress is possible, you can mention this as well)
52nd question by Claude: What are the most cost-effective ways to reduce wild animal suffering that don’t involve large-scale ecosystem disruption?
55th question by Claude: Which animal welfare interventions have the highest suffering-reduction per dollar spent, and which of those also have positive or neutral effects on human interests?
56th question by Claude: Which industries cause significant animal suffering not because it’s economically necessary, but simply due to inertia, tradition, or lack of awareness? What would transition pathways look like?
57th question by Claude: What changes to animal farming practices would significantly reduce animal suffering while having minimal or zero impact on food prices, food security, and farmer livelihoods?
58th question by Claude: What farming innovations (cellular agriculture, precision fermentation, etc.) could become cost: competitive with conventional animal products within 20 years, and what would accelerate them without requiring people to sacrifice anything?
59th question by Claude: What legal rights or protections for animals would most reduce suffering while being politically achievable and not threatening human food security or livelihoods?
60th question by Claude: What do we know about fish and invertebrate sentience, and what low-cost changes to fishing and farming practices would be warranted if we take that evidence seriously?
73rd question by Claude: What can an individual do today to reduce animal suffering that requires no sacrifice to their health, finances, or time?
78th question by Claude What corporate policies reduce animal suffering while also being good for business (through brand value, efficiency, etc.)?
79th question by Claude: What technologies already exist that could reduce animal suffering at scale but are underutilized, and why?
80th question by Claude: What small changes to consumer products (food, cosmetics, clothing) would reduce suffering significantly while being undetectable to most consumers?
Climate change
21st question by GEMINI 3 Pro: If you could change exactly one social norm or habit in the developed world that would have the largest downstream positive effect on climate change (more than electric cars or recycling), what would it be? Trace the causal chain.
Cognitive science
8th question by Zlatko: In which ways human cognition differs from AI cognition?
9th question by Claude: What is consciousness, and do you have it?
Culture
2nd question by Zlatko: Please make a list of 100 greatest songs of all time in languages other than English. Pick songs based on quality and greatness, regardless of whether they are famous internationally. Feel free to include great songs that are not famous outside their regional culture. Try to include songs from all around the world, but the main criteria should be quality.
Economic development
4th question by Zlatko: How much money and time would it take some alliance of developed countries to help the least developed countries in such a way to raise HDI of all the countries to at least 0.5? (So if a country has HDI of 0.49 it would receive very little aid, just to get to 0.5, and from that point it would develop on its own. But if its HDI is 0.38 it would receive much more help. Countries above 0.5 would not receive aid in this project) What would it take politically for such a project to succeed?
20th question by GEMINI 3 Pro: Calculate the cost and logistical requirement to fortify the world’s 5 most common staple foods (rice, wheat, corn, etc.) with the specific micronutrients required to eliminate the 3 most common cognitive-development deficiencies in children globally. What is the ROI on global GDP in 20 years?
88th question by Zlatko What steps should countries make in order to boost their own economic growth and make it sustainable?
Economics
18th question by GEMINI (modified): Find a concept that is considered common knowledge in Physics but is virtually unknown in Economics. Explain how applying this physics concept to economics could revolutionize how we understand markets.
23rd question by GEMINI 3 Pro: Take the 3 most valid criticisms of Capitalism and the 3 most valid criticisms of Socialism. Design a new economic model that specifically addresses these 6 failure points without reverting to the standard arguments of either side.
Education
41st question by Claude: What skills have disproportionate returns that most education systems don’t teach?
Ethics
6th question by Zlatko: Can utilitarianism, deontology and virtue ethics be reconciled? Can a theory that supersedes all of them be developed? What would it look like?
86th question by Zlatko What is the consensus among virtue ethicists when it comes to development of virtues? How to do it in practice?
Existential risk
25th question by GEMINI: (modified): Based on human history and current technological trajectories, what is the most likely ‘Great Filter’ (existential threat) that we are underestimating?
25th question by GEMINI (original): Based on human history and current technological trajectories, what is the most likely ‘Great Filter’ (existential threat) that we are underestimating? Not nuclear war or AI, but something subtler, like fertility collapse or epistemic fragmentation.
72nd question by Claude: What safeguards should be built into any movement for animal or AI rights to prevent it from being captured by ideologies that harm human flourishing?
74th question by Claude: What are the risks of expanding moral circles too quickly, and how can those risks be mitigated while still making progress?
75th question by Claude: How do we prevent concern for non-human welfare from becoming a vector for misanthropy or civilizational risk?
Geopolitics
87th question by Zlatko What principles of foreign policy could a country adopt to behave as a decent and ethical global actor, while also promoting their own interests?
94th question by Zlatko: Is there anything we can do (and who would be the one doing it), to decrease number of wars and conflicts in the world?
Governance
12th question by GEMINI 3 Pro: Analyze the decision-making structures of super-organisms (like ant colonies, mycelial networks, or immune systems). Map these principles onto human governance. How could we design a city council or a corporate board that functions with the decentralized efficiency and resilience of an immune system?
13th question – GEMINI 3 Pro: Identify 5 specific financial or social mechanisms used in the Roman Empire or Han Dynasty that solved a problem we currently have (e.g., inflation, social cohesion), but which we have entirely forgotten or discarded.
Harmonizing interests
66th question by Claude: What does a positive-sum vision of the future look like that includes humans, animals, ecosystems, and potentially AI—where none of these groups are sacrificed for the others?
68th question by Claude What would a constitution look like that protects human rights while also encoding protections for other sentient beings in a stable, non-self-undermining way?
69th question by Claude What metrics beyond GDP or HDI could we use to measure civilizational progress that include non-human welfare?
76th question by Claude: What would a “measured moral expansion” roadmap look like—staged progress that ensures each step is stable before the next is taken?
77th question by Claude: What historical examples exist of societies successfully balancing human interests with obligations to other beings, and what made them stable?
Human welfare
33rd question by Claude: What are the cheapest interventions with the highest quality-of-life improvements that most governments aren’t implementing?
34th question by Claude: Which currently legal policy changes would have the largest positive impact with the least political resistance?
35th question by Claude: What coordination problems, if solved, would unlock the most human potential?
37th question by Claude: If you had $1 billion to improve long-term human flourishing, and couldn’t spend it on any cause that already has major funding, what would you do?
Interdisciplinary research
29th question by Claude: What do neuroscience, economics, and ecology all agree on about complex systems that could inform political design?
30th question by Claude: Which unsolved problems in different fields might actually be the same problem wearing different masks?
36th question by Claude: What are the most important questions that nobody is seriously researching because they fall between disciplines?
Life advice
3rd question by Zlatko: Please write a list of 100 actionable facts, that if people knew could greatly improve their lives and the world. Don’t write tips or advice, just facts.
5th question by Zlatko: Make a list of 10 most useless and dumb aspirations people have, that should better be avoided?
22nd question by GEMINI 3 PRO: List 20 physical items or tools costing under $50 that save the most amount of time or prevent the most amount of physical frustration over a 10-year period. Focus on ‘buy it for life’ utility.
32nd question by Claude: If you had to compress all of human wisdom about living well into 10 principles that don’t contradict each other, what would they be?
40th question by Claude: What questions should a person ask themselves once a year that most people never ask?
46th question by Claude: What questions am I not asking that I should be asking
82nd question by Zlatko: What advice could you give to individuals to improve their relationships with their family members?
83rd question by Zlatko: What advice could you give to individuals to improve their social life?
84th question by Zlatko What changes could one make to significantly improve one’s quality of life?
85th question by Zlatko: What is the best way to improve one’s personality?
Linguistics
45th question by CLAUDE: What concepts don’t exist in English but should, based on concepts in other languages or gaps in our thinking?
Medical research
89th question by Zlatko: What blindspots do we have when it comes to cancer research? What changes in research direction could accelerate finding a cure?
90th question by Zlatko What blindspots do we have when it comes to Alzheimer’s research? What changes in research direction could accelerate finding a cure?
91st question by Zlatko: What blindspots do we have when it comes to multiple sclerosis research? What changes in research direction could accelerate finding a cure?
92nd question by Zlatko What blindspots do we have when it comes to ASL research? What changes in research direction could accelerate finding a cure?
93rd question by Zlatko: What diseases are most neglected and underresearched? Please apply Importance, Neglectedness, Tractability to medical research? Which diseases should we research more?
95th question by Zlatko: What should we do to avoid antibiotic resistence disaster? Is there anything we can do besides just being more careful and restrained while prescribing antibiotics?
96th question by Zlatko: What should we be doing to prevent future pandemics?
Meta
1st question by Zlatko: Can you imagine what would be 50 greatest prompts of all time that people can make and send to AIs?
Non human welfare
67th question by Claude: What economic or political systems might better account for the interests of beings that cannot vote, pay, or advocate for themselves?
70th question by Claude: What is the strongest philosophical case for expanding moral consideration to non-human entities that doesn’t rely on controversial metaethical assumptions?
71st question by Claude: What would a practical ethical decision procedure look like for individuals, companies, and governments that takes non-human welfare seriously without being self-defeating?
81st question by Claude: What research questions, if answered, would most clarify our obligations to non-human entities?
Philosophy
7th question by Zlatko: What is more important, pleasure of flourishing?
10th question by Claude: What would you ask humanity if you could ask one question?
11th question by Claude: Steelman the position you find least defensible
19th question by Gemini: What is a philosophical concept or emotional state that is central to Eastern (or Indigenous) thought, has a specific word for it in those languages, but is completely absent from the Western conceptual framework? How would adopting this concept solve a specific Western mental health crisis?
31st question by Claude: What would a synthesis of Eastern and Western philosophical traditions look like if neither side had to “win”?
44th question by Claude: What are we almost certainly wrong about today that will seem obvious to people in 100 years?
Rationality
14th question – by Gemini 3 Pro: List 20 common modern interventions (in medicine, parenting, or policy) where the ‘cure’ is statistically worse than the disease, based on long-term data. Focus on areas where ‘doing nothing’ yields better outcomes.
15th question – by GEMINI 3 Pro: Identify the 10 specific linguistic patterns or rhetorical tricks used in modern media that degrade human cognitive processing and attention span. Don’t just list logical fallacies; list formats (e.g., infinite scroll, rage-bait headlines) that physically alter how we think.
16th question – by GEMINI 3 Pro: List 50 concepts that are widely taught in schools or business books that have been scientifically debunked or are statistically irrelevant, yet persist due to tradition. Help me unlearn them.
24th question by GEMINI: Write a report from the perspective of an alien anthropologist observing humanity. Describe our strangest ritual that we consider normal, but which actually causes us the most unnecessary suffering.
38th question by Claude What mental models exist that, once learned, permanently upgrade decision-making across all domains?
39th question by Claude What are the most common reasoning errors that intelligent people make specifically because they’re intelligent?
42nd question by Claude 33: What would a checklist look like for making important life decisions that incorporates findings from behavioral economics, philosophy, and psychology?
43rd question by CLAUDE: What false dichotomies structure most public discourse, and what would more accurate framings look like?
Science
17th question by GEMINI: Scan through scientific literature to identify 5 discoveries or technologies that showed immense promise in peer-reviewed papers 20+ years ago but were never commercialized or adopted due to timing or lack of funding. Which ones are viable today due to technological advancements?
26th question by Claude: What are the most important findings in non-English academic literature that haven’t crossed into English-language discourse?
27th question by Claude: What are the most robust findings in social science that journalists almost never report on?
Technology
28th question by Claude: What technologies exist right now that could be widely deployed but aren’t, for non-technical reasons?