{"id":158759,"date":"2026-09-12T13:07:21","date_gmt":"2026-09-12T17:07:21","guid":{"rendered":"https:\/\/marcstier.com\/blog2\/?p=158759"},"modified":"2026-10-08T13:18:14","modified_gmt":"2026-10-08T17:18:14","slug":"ai-as-a-useful-tool","status":"publish","type":"post","link":"https:\/\/marcstier.com\/blog2\/?p=158759","title":{"rendered":"AI as a Useful Tool"},"content":{"rendered":"

My Skepticism About AI<\/strong><\/p>\n

If you have been reading my posts about AI on Facebook over the last year, you know that I\u2019m something of an AI skeptic. For a few reasons, I have very real doubts about the ability of AI, as it currently exists, to do really creative work on its own because creative work is impossible without the human drive to find certain things out or solve a particular problem that few have been concerned about in the past. And what we take to be a problem worth solving is often the result of our fairly idiosyncratic takes on the world that come from our unique experience.<\/p>\n

AI Might Lead to Artificial Life\u2026Some Time From Now<\/strong><\/p>\n

I do think AI could be developed into a life form that would have unique experiences that would lead to an idiosyncratic perspective that in turn generates desires to solve new problems. But that would most likely involve creating an AI that is embodied in machines that can interact with the world in sensual and motoric ways, that have a few fundamental drives along the lines of the basic human drives that motivate us that underlie its actions, that is trained in a particular culture over time, and that has a much longer existence than AI agents have today. My guess\u2014which I will defend in more detail at some point\u2014is that only this kind of artificial life is likely to develop the idiosyncratic desires beyond those which we give AI that will generate true creativity. Until that happens, the creativity of AI will be parasitic on the creativity of the human beings who give it its purposes.<\/p>\n

The kind of artificial life I\u2019m discussing is far beyond what AI is today, although someone who realized why these things are important and sought to create such an artificial life form might be able to create it soon, perhaps even in the next twenty or fifty years.<\/p>\n

AI Is, For Now, a Very Useful Tool<\/strong><\/p>\n

\"\"But right now, AI is far more limited. AI is mostly designed to solve the problems we give it. That does not mean AI cannot come up with creative ideas. But its creativity is a derived one, that is based on our own creativity in asking it to solve certain problems.<\/p>\n

That, however, does give AI an opportunity to be extremely helpful to us. It is a tool that, in a variety of ways, can make our thinking more effective and faster. That\u2019s true for four reasons. Two of them most people know about. AI is fast. And properly trained, it has access to a huge amount of information that it can combine in new ways. The other two reasons are very different in nature and have to do not with AI\u2019s abilities but its nature as a tool. I\u2019ll return to them after reviewing two AI experiences I had recently.<\/p>\n

An Example: AI and Work Planning <\/strong><\/p>\n

A few weeks ago, I saw an AI-related job search website that I thought might be able to generate some prospects for the consulting firm I\u2019m planning to get up and running in about a month. The website had an interview process run by an AI agent that interacted through voice. The process was designed to find out what a potential job applicant was interested in doing and then suggest potential jobs.<\/p>\n

I spent about 15 minutes on this process discussing the political advocacy and organizing work I have done. I gather that this was longer than usual. And the reason it was longer than usual is that as I described work I had done in the past, the AI agent asked some very good questions that helped me spell out my ideas more clearly and talk about how two aspects of my work, the policy research and analysis I did on the one hand, and my organizing and advocacy work on the other, related to one another. I didn\u2019t come up with new ideas about how those two kinds of work were interrelated through this process. It was something I had thought about and articulated before. But thanks to the agent\u2019s questions and my effort to explain to the agent what I mean in terms it could understand, I found some new ways of formulating my ideas.<\/p>\n

And then the agent asked me how I might use AI to do the organizing \/ advocacy work better or more efficiently. I had thought about this a bit, but I went a lot deeper in developing ideas in response to the agent\u2019s questions. And its summary of what I said and follow-up questions led me to come up with other ideas about using AI in advocacy and organizing that I had not thought of before.<\/p>\n

I was the one coming up with some new ideas in this process, not the AI agent. But the conversation with the AI agent, and the questions it asked, helped me do so. So what the agent contributed here was in being an effective thought partner in my own intellectual process.<\/p>\n

Did the AI agent do something that a human thought partner might not have? No and yes. It asked the same kinds of questions a human partner might have asked. What AI gave me was two things:<\/p>\n

AI as an Endlessly Patient Thought Partner<\/strong><\/p>\n

The first was a willingness to talk about my ideas. Most of us, if we are lucky, do have thought partners willing to engage in brainstorming ideas. But frankly, we often don\u2019t have them as often as we would like. And in the world of work, there are pressures to limit the kind of brainstorming that really helps our own creativity. At PPC I would sometimes use staff meeting time to do some brainstorming about our larger strategies. I did this both to reflect on our work and to generate better ideas. And I did it so that the staff could understand and evaluate the strategies I proposed in more depth, with the hope that they would become better at strategic thinking themselves.<\/p>\n

The Virtue of Brainstorming <\/strong><\/p>\n

Some members of the staff seemed to enjoy these conversations. But others, including our deputy director for operations,were impatient. They wanted to leave the staff meeting or other meetings with a short list of things they had to do, not with a better understanding of our political strategy. \u00a0And they wanted staff meetings that ended in the aloted time. This, in my view, was shortsighted. I believed that if the staff had a better understanding of our strategy as I understood it, and a better grasp of strategic thinking, over the long term they would become more creative at doing their work. And they would need less supervision from me. The four members of the staff who did appreciate these brainstorming sessions became much more creative and more confident in their work over time. They also made my life easier. The two who seemed bored or annoyed by these sessions and were impatient for them to end mostly did not.<\/p>\n

So one advantage of AI as a thought partner is that it has unlimited time and patience for trying out ideas. It is available 24\/7.<\/p>\n

Brainstorming without Self-Esteem Issues <\/strong><\/p>\n

When one is brainstorming, one often finds that ego gets in the way. Someone will come up with an idea and not let it go even as others demonstrate its flaws. Or new ideas are shot down because they challenge preconceived notions. Or because they might require staff members to develop new ways of working or do things that they find uncomfortable instead of just doing what they always do.<\/p>\n

AI agents, at least so far, don\u2019t have self-esteem issues of this kind. And they don\u2019t have pride of ownership. They are willing to entertain ideas they have never heard and see where they go. That can make brainstorming very effective.<\/p>\n

AI was useful then not just because it thought fast and could draw on a range of information, but because it had time, patience, and no ego.<\/p>\n

Another Example: Financial Planning <\/strong><\/p>\n

The next day I got some new financial planning software. My wife and I are retiring. We have a few different pots of retirement funds of different kinds. We have to make some choices about which funds to draw on first and also whether to convert one kind of pot into another. We need to decide such things as whether we should buy an annuity or convert a traditional IRA into a Roth IRA, etc. And in running out these various scenarios we need to think about different rates of returns and how they might vary under different conditions, tax rates on different kinds of funds and how they might change, the impact of drawing down some funds on health care costs thanks to IRMAA, and so forth.<\/p>\n

Earlier in the week I had developed a fairly large Excel spreadsheet model that projected future earnings and wealth and our taxes based on variations in our decisions. It took five or so hours to build over two days. And it was a little cumbersome to use.<\/p>\n

It occurred to me that there might be financial planning software that could do all this work. Now I was glad I had built my own model before looking for such software because building and using my own model taught me a great deal about all the factors involved in a financial plan and how they interact. But I was hoping I could find something even more thorough than my own model and that could check my work.<\/p>\n

I found a highly recommended online software tool called Boldin. I set it up and found first that it confirmed that my Excel spreadsheet was pretty accurate. And I also found that Boldin allowed me to work faster in trying different scenarios than I could do with my Excel spreadsheet.<\/p>\n

And then I discovered that it had a built-in AI help system. So I started experimenting.<\/p>\n

The first thing it did was explain to me that an odd outcome I was getting was likely the result of two accounts being set up incorrectly. It showed me how to change those accounts and offered to do it for me. I did not take up the offer both because I wanted to learn the software and because I was about half done by the time it gave me the last procedure.<\/p>\n

Then things got more interesting. I started using it to investigate the best way to withdraw funds from one of my IRAs. I had thought of doing so in even amounts over a few years. It suggested instead that I should delay withdrawals until the end of the period to allow for a longer period of tax-free buildup of assets. This isn\u2019t a brilliant but rather a commonplace idea. But the AI system did propose it to me before I thought of it. I asked it to give me the results of both approaches and it came back with an answer in 60 seconds showing that its suggestion made sense. Then I asked it to modify the plan to allow for an earlier withdrawal of enough money to fund a home renovation project. Again, 60 seconds later I had a plan to do that along with an analysis of the financial implications.<\/p>\n

And then things got really interesting. The financial plan it had come up with basically drew down funds from our four pockets of money roughly equally. But I had an idea that I gather was unusual. Having built the Excel model, I knew our accounts well enough to wonder whether we might get substantial tax savings not just for us but our daughter by taking most of the money from three of the four funds while not touching the last one. I suggested that the AI agent model this idea.<\/p>\n

It told me that this was a very unusual approach but that it would help me figure out if it made sense. It took about 2 minutes for it to do the calculation and provide me with a chart showing withdrawals of the kind I specified as well as an analysis of the result.<\/p>\n

It turned out that my idea saved us a lot of money. (By a lot, I mean a lot relative to what we have, not a lot in absolute terms!)<\/p>\n

I stopped there because I had other stuff to do and will return to this project soon.<\/p>\n

So what did this experience tell me?<\/p>\n

AI Works Fast<\/strong><\/p>\n

First, that AI models trained to do a certain kind of work can be very helpful. AI can develop a financial projection that accounts for all the relevant factors that financial plans usually consider. Because it knows the kinds of moves financial planners often make, it can suggest ideas that make sense, like delaying withdrawals from one of my accounts.<\/p>\n

AI Has a Deep Knowledge Base<\/strong><\/p>\n

Second, just as I experienced in my other AI interaction, AI was really helpful. The Boldin agent could draw on a substantial amount of information about the interaction of the different factors that shape financial plans for retirees. It could do calculations very quickly and present them in effective ways.<\/p>\n

AI Has Infinite Patience<\/strong><\/p>\n

All these AI abilities are well known. But what I most appreciated was that the Boldin agent was willing to help me test my own ideas. It had infinite patience for trying those ideas. It had no ego problems in my suggestion that it make a huge change in the financial plan it had developed for me. And it was not time constrained.<\/p>\n

So again, AI turned out to be a truly useful tool. It\u2019s not a replacement for a human being. The most creative idea developed in this process of financial planning did not come from the Boldin AI agent but from me. But it was extremely helpful in testing that idea. And I think it will be even more helpful as I refine the plan I developed as I suspect it will have suggestions about how to implement it well.<\/p>\n

When I first started looking at AI I concluded that it was going to be a great tool for people who do serious thinking but that it would not replace them \/ us. And I thought that AI could be part of an intelligent life form that could do what we do although I suspected this was 100 years away.<\/p>\n

I now see ways in which that human made intelligent life form could come into existence much more quickly. But I think we are a long way away. In the meantime, while I\u2019m still pessimistic about the notion that AI is going to eliminate all human work and take all our jobs, I\u2019m more optimistic about it being a helpful tool for human beings doing work that involves the production of ideas. I can think of many ways that AI could have helped me do the work I did at PPC. And I\u2019m really grateful to be alive at a time to take part in the human effort to use AI tools to become much more productive than we have been in the past.<\/p>\n","protected":false},"excerpt":{"rendered":"

My Skepticism About AI If you have been reading my posts about AI on Facebook over the last year, you know that I\u2019m something of an AI skeptic. For a few reasons, I have very real doubts about the ability of AI, as it currently exists, to do really creative work on its own because creative work is impossible without the human drive to find certain things out or solve a particular problem that few have been concerned about in the past. And what we take to be a problem worth solving is often the result of our fairly idiosyncratic takes on the world that come from our unique experience. AI Might Lead to Artificial Life\u2026Some Time From Now I do think AI could be developed into a life form that would have unique experiences that would lead to an idiosyncratic perspective that in turn generates desires to solve new\u2026 Continue reading<\/a><\/p>\n","protected":false},"author":1896,"featured_media":158760,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"{title}\n\n{excerpt}\n\n{url}","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2},"_wpas_customize_per_network":false,"jetpack_post_was_ever_published":false},"categories":[319],"tags":[],"class_list":["post-158759","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"jetpack_publicize_connections":[],"jetpack_shortlink":"https:\/\/wp.me\/p35YuU-FiD","jetpack_sharing_enabled":true,"jetpack_likes_enabled":true,"jetpack_featured_media_url":"https:\/\/i0.wp.com\/marcstier.com\/blog2\/wp-content\/uploads\/2026\/10\/image_b309b98b.jpg?fit=1408%2C768&ssl=1","_links":{"self":[{"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/posts\/158759","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/users\/1896"}],"replies":[{"embeddable":true,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=158759"}],"version-history":[{"count":1,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/posts\/158759\/revisions"}],"predecessor-version":[{"id":158761,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/posts\/158759\/revisions\/158761"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=\/wp\/v2\/media\/158760"}],"wp:attachment":[{"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=158759"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=158759"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/marcstier.com\/blog2\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=158759"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}