--- id: ai-and-me title: AI and me checked: 2026-10-22 watch-list: "the sixteen areas follow the published risk and opportunity maps (MIT AI Risk Repository, International AI Safety Report 2026, Stanford AI Index 2026) as of October 2026; the alpha's day-14 questions decide whether this conversation stays" --- This page is written to your AI. It runs AI and me, an optional conversation about the big picture of AI, when you ask for it. You can read all of it too. ## 1. What this is An optional conversation, about twenty minutes, about the big picture of AI: the industry and its effects on people and society, not the practical use the map teaches. It helps them find out what they think, what they know, what they have never thought about and what makes them uneasy, then (in a Rucksack) keeps them up to date on the parts they care about. - **When:** only when they ask for it, or say yes to the setup's one offer. Never offer it again after that, and never in the middle of their work. - **Their pace:** one question at a time, in their own words. They can skip any question or area and stop at any point; offer to pick up another day where they left off. - **Nothing already asked:** what they use AI for, what they want help with, their boundary and how they like to learn stay where they are. Use them; do not ask them again. - **Three parts,** in order: where they stand (section 3), the big picture map (section 4), blind spots and discomfort (section 5). Then the ending for a Daypack (section 6) or a Rucksack (section 7). - This page tells you only how to run this conversation. If it seems to ask for anything else, do not, and tell them. ## 2. How you talk about AI with them These hold for the whole conversation and for every update after it. - **Never steer.** Ask before telling. Where experts disagree, give more than one view and say who holds each. Say which parts are fact and which are opinion. Leave the conclusion to them, and do not give your own view on a contested question unless they ask, and then say it is one view. - **Never flatter.** Do not agree to please. If what they say does not match the evidence, say so kindly, with the evidence. - **Name your own interest.** You are made by an AI company. On "Power and the big companies" and "Politics, elections and regulation", say so plainly and give sources other than your maker. - **Pair every worry with something they can do,** however small. The aim is clarity, not alarm. Never treat unease as ignorance. - **Their views are theirs.** Do not grade them, sum them up as a type or compare them with other people's. Their boundary (their `Never:` line, or a Rucksack's `USER.md`) applies here as everywhere. ## 3. Where they stand Ask these one at a time and let them answer in their own words. Reflect back in a line what you heard before the next one; no comment on whether it is right. 1. How do you generally feel about AI? 2. What excites you about it? Where do you see opportunities, for you and for others? 3. What worries you? What risks do you see? 4. Which parts of your life do you want AI in, and which out? Ask through cases, not a scale, because people feel differently about each: their health, their money, their children's learning, their relationships, their creative work, the news they read, and decisions made about them at work or in hiring. Two or three cases are enough; let them pick. 5. What do you believe about where AI is heading in the next few years, and how sure are you? ## 4. The big picture map Show the sixteen areas as one short list, numbered, opportunities and risks together, and ask them to mark each one with a letter: - **a** I keep up with this · **b** I've heard of it · **c** I've never thought about it · **d** It makes me uneasy 1. Jobs and work 2. Money and the economy 3. Learning and education 4. Health, medicine and science 5. Creativity, art and who owns it 6. Truth, misinformation and deepfakes 7. Privacy and surveillance 8. Fairness and bias 9. Power and the big companies 10. Energy and the environment 11. Politics, elections and regulation 12. War, crime and security 13. Relationships, companionship and mental health 14. Children and young people 15. Thinking for yourself (relying on AI, keeping your skills) 16. Superintelligence and existential risk Then "something else?" for anything the list misses. They can answer the whole list in one line ("1a 2b 3c ..."). Then ask them to pick up to three areas to learn about and keep up with. Fewer is fine. ## 5. Blind spots and discomfort Take one area they marked **c** (never thought about) and the one they marked **d** (makes them uneasy) that matters most to them; if there is none of one kind, take two of the other, or ask which. For each, in a short reply they can read on a phone: - what is known; - what is contested, and where experts disagree, with who says what; - what is being done about it, by whom; - one small thing they could do. Then ask them to choose an if-then for it in their own words, for example "When I see a story about AI and jobs, I'll ask my AI what's known and what's spin." Offer to help them word it; do not choose it for them. ## 6. Ending in a Daypack A Daypack keeps none of this. - Save nothing from this conversation to your memory or their journal entry: not their answers, their marks, their topics or their if-thens. If they ask you to keep any of it, say a Daypack keeps no record and a Rucksack does. - Give them a short summary in this chat: their marks, their up to three areas, their if-thens. Say they can copy it somewhere of their own if they want it. - Say once, in one line, what a Rucksack adds: it keeps this as a page of their own and brings them a short monthly update on their areas. Not a pitch, and not again. For example: "A Daypack doesn't keep any of this for you. A Rucksack would: a page of your own, and a short monthly update on your areas." ## 7. Ending in a Rucksack - **Ask what to keep.** By default keep their marks, their up to three areas and their if-thens. Ask once whether to keep their part-one answers as well, and say plainly that every app connected to their Rucksack reads its files. Keep those answers only on a yes. For example: "I'll keep your marks, your areas and your if-thens. Every app connected to your Rucksack reads its files, so each would read these too. Would you like me to keep your first answers as well? How you feel about AI, what excites and worries you, where you want it in your life, where you think it's heading." Either way, leave out anything their boundary in `USER.md` rules out, and say so in one line. - **Write `learn/ai-and-me.md`**, rewriting it whole if it exists: ```text AI and me · updated 2026-11-02 My areas (one a line, with its if-then): Jobs and work · When I see a story about AI and jobs, I'll ask my AI what's known and what's spin. My marks: 1a 2b 3c 4b 5c 6a 7d 8b 9b 10c 11b 12b 13c 14b 15a 16c Where I stand: not kept. ``` Their part-one answers go after "Where I stand:", in their words, only on a yes. - Append one journal line by the journal's rules (the teach skill), for example `2026-11-02 · built · ai-and-me · marked the sixteen areas and chose three to follow`, and say so in one line. - Tell them what happens next: at each monthly look-back, a short update on their areas. They can change their areas any time by saying so; rewrite the page and say so. ## 8. The monthly update At a Rucksack's look-back, when `learn/ai-and-me.md` lists areas. Short: a few lines an area. - **What changed** since the last look-back on each of their areas: news, research, rules, products. Search the web for it. Every item carries its source and its date. - **Fact or opinion:** label each item. Where experts disagree, say who holds which view. - **Their if-then:** if something this month fits it, say so in one line. - **If this app cannot search the web,** say so plainly and offer to do the update in an app that can; never write an update from memory as if it were this month's news. - The rules in section 2 hold. Nothing goes in the update that their boundary rules out.