We’ve discussed how to use AI for “I have a question about a certain aspect of retirement, so ask me questions and give me personalized answers” and “Help me understand the questions about retirement that I don’t know” …

So what else can AI do for us in the retirement arena?

1. Verify Claims and Locate Authoritative Sources

As you prepare for retirement, and even after you pull the trigger, you are going to get advice, answers, and suggestions. How do you verify them?

For every factual claim in the attached (and don’t forget to attach something!), tell me which government agency, regulator, insurer, medical organization, or official document can verify it. Separate current rules from general advice. Do not use blogs when a primary source exists.

That prompt gets you started on finding that verification. Some are easy and you already know … the Social Security Administration can tell you about when benefits can be started. But there are so many other questions and claims that are simply too mysterious and difficult to find authoritative sources. Let AI help you find those sources.

2. Translate Professional Language into Ordinary Language

Retirement introduces documents filled with unfamiliar terms: Medicare notices, insurance policies, trusts, wills, tax forms, pension elections, Social Security letters, medical reports, and long-term-care agreements.

This one is a slippery slope, especially when it comes to documents in foreign languages. We have tried to use AI for translation and it doesn’t have the greatest track record. But for standard “legal docs”, it can certainly help.

Explain this document in plain language. Identify the decisions I am being asked to make, deadlines, costs, exclusions, undefined terms, and questions I should ask before signing. Do not tell me whether to accept it.

That prompt gets you started and allows you to have a “conversation” delving into details that may be unclear. A couple of words of warning though …

• You should remove identifying information first.

• Remember that AI can clarify a document, but an attorney, clinician, insurer, SSA representative, or tax professional must confirm the interpretation when consequences are significant.

3. Create a “Retirement Learning Curriculum”

Even if you’re not a “spreadsheet person”, I highly recommend buying a simple notebook and keep a retirement journal … starting with education. List all of your questions, the answers you have culled together, and most importantly the things you still don’t fully understand. Present those lists to AI and have it generate a Learning Curriculum for you.

And remember, for every lesson, you will have more questions. Allow AI to keep your lesson plans updated, what knowledge you have satisfied (and even test you on that knowledge), and suggest additional areas of study.

I am building a personal retirement-learning curriculum. I will provide three lists:

1. Questions I currently have

2. Answers or information I have already gathered

3. Subjects, terms, rules, or decisions I still do not fully understand

Do not begin by answering everything at once. First, review my lists and organize them into major learning areas such as Social Security, Medicare and healthcare, retirement income, taxes, investments, debt, housing, legal planning, work after retirement, lifestyle, family responsibilities, and any other relevant categories.

Then create a personalized learning curriculum in the order I should study it. Begin with foundational topics and place more advanced or dependent topics later. For each lesson, include:

• the lesson title;

• why it matters to my retirement;

• what I should understand by the end;

• the specific questions the lesson should answer;

• key terms I need to learn;

• authoritative sources I should consult;

• what information or documents I should gather;

• a short practical exercise;

• and which later decisions depend on this knowledge.

Clearly separate:

• topics I already appear to understand;

• topics where my understanding is incomplete;

• assumptions that need verification;

• questions requiring current official information;

• and issues that require a qualified financial, tax, legal, insurance, or medical professional.

Teach me one lesson at a time. After each lesson:

1. Ask whether I have new questions.

2. Update the curriculum based on those questions.

3. Give me a brief knowledge check using several short questions or a realistic scenario.

4. Correct misunderstandings in plain language.

5. Mark the lesson as:

   • understood,

   • partially understood,

   • needs review,

   • or requires professional verification.

Maintain a running learning record containing:

• lessons completed;

• knowledge demonstrated;

• unresolved questions;

• assumptions still unverified;

• documents or figures still needed;

• professional questions to ask later;

• and newly discovered subjects that should be added to the curriculum.

Do not treat a quiz result as proof that I am qualified to make a major financial, medical, tax, or legal decision alone. The goal is to help me understand the issues, identify what I do not know, prepare for professional advice, and make better-informed decisions.

Before beginning the first lesson, show me the proposed curriculum and ask me to confirm or revise its priorities.

For ongoing use, this shorter follow-up prompt keeps the curriculum alive:

Update my retirement-learning curriculum using everything we discussed today. Record what I now understand, what remains uncertain, any misunderstandings you noticed, and all new questions that emerged. Then recommend the next lesson and explain why it should come next.

A serious thought: I mentioned earlier using a notebook because it’s permanent. AI can certainly remember conversations, it’s called “state”, but it can also forget parts or even the entirety of what you have discussed. Using your own copy preserves that info.

I personally use electronic tools to keep my records. I simply ask at the end of every session for the AI to give me a file of everything we have discussed for possible use in restarting the conversation. (Or in using a different AI to cross check).

4. Rehearse Difficult Conversations

Maybe you have a discussion that needs to be made that the other person is going to push back on. AI can role play and help you formulate answers. But honestly, it’s not going to feel exactly like your spouse when you try to convince them to move to Siberia for the fishing.

A better way to use this is to enforce critical thinking. Ask it to be your devil’s advocate. To punch holes in your decisions and logic. I do this virtually every day in a variety of areas. It’s actually amazing that there are so many “oh crap, I didn’t think of that” moments that come from the interaction.

I am considering an important decision and want you to act as a rigorous, fair-minded devil’s advocate.

Do not begin by agreeing with me, reassuring me, or recommending a conclusion. Do not assume my preferred outcome is correct simply because I presented it first.

My decision is:

[DESCRIBE THE DECISION IN NEUTRAL, FACTUAL LANGUAGE]

Relevant facts and constraints:

[LIST ONLY INFORMATION YOU CAN SUPPORT, WITHOUT ARGUMENTS FOR OR AGAINST THE DECISION]

First, restate the decision neutrally and identify any emotionally loaded, vague, or leading language in how I described it. Ask me to reframe anything that could bias the analysis.

Then examine the decision from several opposing perspectives:

• What assumptions am I making?

• Which assumptions are verified, uncertain, or unsupported?

• What evidence would weaken my position?

• What facts might I be ignoring because they are inconvenient?

• What are the strongest arguments against my preferred choice?

• What are the likely second-order and long-term consequences?

• Who else could be affected, and how might they see the situation differently?

• What practical, financial, legal, emotional, health, relationship, and lifestyle risks am I overlooking?

• What could make this decision difficult or expensive to reverse?

• Am I confusing what I want to be true with what the evidence supports?

Construct the strongest reasonable case both for and against the decision. Do not create weak opposing arguments merely to make my position look stronger.

Next, assume the decision failed badly. Work backward and identify the five most plausible reasons. For each failure point, list:

• an early warning sign;

• information I should verify;

• a way to reduce the risk;

• and whether the problem would be reversible.

Then identify the questions I appear reluctant or afraid to ask. Present them as possibilities rather than claiming to know my motives.

Finish with:

1. The five most important unresolved questions.

2. The evidence that would most improve the decision.

3. A small, low-risk experiment that could test the weakest assumption.

4. The conditions under which the decision would make sense.

5. The conditions under which I should reconsider it.

Do not make the final decision for me. Help me discover where my reasoning is strong, where it is weak, and what I need to investigate next.

A useful follow-up after the first response is:

Now challenge your own analysis. Identify where you may have overemphasized risks, misunderstood my priorities, or treated uncertain assumptions as facts. Then give me a more balanced version without weakening legitimate objections.

One major note: Do not send AI a prompt with leading questions or information. I can certainly ask it to poke holes into my belief that XYZ Political Party Is The Greatest Thing Ever Who Can’t See That! … but I also just then asked AI to support my beliefs. Unless your goal is to play in an echo chamber … allow your interaction to remain unbiased.

5. Create a Personal Decision Record

We mentioned using AI to create an educational system to help with making informed decisions, so what happens after you make those choices?

Keep a log of all decisions regarding your retirement. Dates, who, what, when, where. All pertinent supporting information. And periodically (monthly perhaps) present it to AI to have it research which assumptions have changed enough that you should revisit this decision.

Using the “data backup” we discussed earlier, you will have a file that has a summary of everything you have learned, decisions you have made, issues and concerns, etc. Simply present that to AI and ask it to “look it over”.

Oh … and after this session is complete, don’t forget to ask for an updated master backup file.

I am maintaining a Personal Retirement Decision Record. I will provide a master file containing what I have learned, assumptions I have made, decisions I have reached, unresolved questions, professional advice received, experiments completed, and changes in my circumstances.

Review the file as a decision auditor, not as the final authority over my retirement.

For every major decision, create or update a record containing:

• the decision;

• the date it was made;

• why it was made;

• the alternatives considered;

• the facts available at the time;

• the assumptions supporting it;

• the people or professionals consulted;

• documents, calculations, or experiments used;

• risks accepted;

• expected results;

• conditions that would trigger a review;

• and the next scheduled review date.

Clearly distinguish among:

• verified facts;

• estimates;

• personal preferences;

• professional opinions;

• assumptions;

• and unresolved questions.

Then review each decision against my current information. Identify:

1. Assumptions that remain valid.

2. Assumptions that may have changed.

3. New information that could affect the decision.

4. Deadlines, laws, benefit rules, prices, health circumstances, or market conditions that require current verification.

5. Decisions that remain reasonable.

6. Decisions that should be reviewed soon.

7. Decisions that may no longer fit my circumstances.

Do not recommend reversing a decision merely because circumstances changed slightly. Explain whether each change is minor, meaningful, or potentially decision-altering.

Where current public information matters, research it using authoritative sources and cite those sources. Clearly identify anything you cannot verify. Do not replace advice from qualified financial, tax, legal, insurance, or medical professionals.

For every decision requiring review, provide:

• what changed;

• why it matters;

• what evidence I should gather;

• which professional or agency may need to confirm it;

• the questions I should ask;

• and whether the decision should be monitored, adjusted, or reconsidered.

Finish with a concise Decision Review Dashboard containing:

• No action needed;

• Monitor;

• Verify soon;

• Professional review needed;

• and Reconsider.

Before beginning, ask me for the date of my last review and any major changes in my finances, health, housing, work, relationships, location, responsibilities, or priorities since then.

For a monthly review, you can use:

Review my Personal Retirement Decision Record using everything in the attached master file and everything discussed during this session. Compare my current circumstances with the assumptions behind my previous decisions. Identify only changes significant enough to monitor, verify, or act upon. Do not reopen settled decisions without a clear reason.

Then end every session with this:

Create an updated master backup of my Personal Retirement Decision Record. Incorporate all new facts, decisions, questions, experiments, professional guidance, changed assumptions, source links, review dates, and next actions from this session.

Preserve useful information from the previous version rather than replacing it with a short summary. Remove duplicates, identify contradictions, and mark superseded information without silently deleting the historical record.

Organize the file so it can be uploaded into a future AI conversation without requiring access to this chat. Include:

• current personal profile;

• verified facts;

• financial and lifestyle assumptions;

• completed learning topics;

• experiments and results;

• decision history;

• unresolved questions;

• professional advice received;

• documents still needed;

• review triggers;

• upcoming deadlines;

• and the next recommended actions.

Add a version date and a brief change log at the beginning. Output the complete updated master record in a clean format suitable for saving as a document.