AI Tools for Women: Practical Uses and Lessons
Discover how the Aged Intelligence Society helps women use AI for caregiving, admin and work—and how to test tools without outsourcing your judgment.
How Women Are Making AI Useful in Midlife
AI adoption is often sold like a race: learn the newest tool, automate everything, or fall behind. But that framing skips a simpler, more urgent question: what would you genuinely like to spend less time doing?
That's where the Aged Intelligence Society comes in. Founded by former editors Shoshana Berger and Nina Willdorf, it runs invite-only gatherings where women test AI tools on everyday, real-world problems - like working through a long-term-care insurance packet or drafting a consulting proposal. The goal isn't to turn people into AI experts. It's to develop enough fluency to tell when these systems help - and when they don't.
That distinction matters. AI can produce convincing nonsense, mishandle sensitive information, and create extra work when you adopt it without a clear reason. Used carefully, though, it can take the edge off repetitive administrative tasks that quietly steal attention. This article looks at how the Society's format works, what its examples suggest about practical adoption, and how you can test similar use cases without outsourcing your judgment.
Why the Aged Intelligence Society started
Berger and Willdorf say the idea took shape during a conversation at a birthday party. They were discussing Claude, an AI assistant, while other women nearby didn't have the context to join in. Berger noticed a gap: smart, accomplished people were missing a chance to test a technology that was already starting to affect work - and daily life.
The founders aren't positioning themselves as AI researchers or cheerleaders for every new product. Their argument is straightforward: the tools are already here, and opting out entirely can leave you less prepared to make good decisions about them. What people need, in their view, is practical fluency - not the belief that every AI feature is automatically useful or safe.
That distinction also fits businesses. Adoption doesn't have to start with a big corporate mandate or a long list of tools. It can start with one task, one test, and an honest look at what happened. The question isn't "How do we use AI everywhere?" It's "Which recurring problem is worth testing, and what would count as a good outcome?"
Why focus on women over 40?
The Society focuses on women over 40 because its founders see a strong fit between this group's experience and the way today's AI tools tend to work. In short: years of professional and personal experience can help people ask sharper questions, spot weak answers, and push a tool to improve rather than just accept what it outputs.
There's also the day-to-day reality. Many midlife women balance work with care responsibilities for children and aging parents. People call it the "sandwich generation," though Berger says she dislikes the term. Whatever you call it, the workload is real: coordination, research, paperwork, and decision-making often land on one person.
AI doesn't fix the underlying structural problems behind that workload. It can't create more caregiving support, make an insurer fair, or rebalance domestic labor. But it can help with specific, information-heavy tasks - especially when the alternative is hours spent sorting, summarizing, or drafting.
The founders' point is personal and economic. If administrative labor cuts into paid work, business development, or rest, reducing some of it can matter. And that "value" should be judged by outcomes, not by how many AI tools someone has installed.
A format built around real problems
The Society's events are intentionally small - usually eight to 12 attendees. That size is meant to keep the room usable. People from different backgrounds can spark useful perspectives, but larger groups can make hands-on experimentation harder to support.
Attendees bring a laptop and a "midlife quandary": a specific problem they want to explore. The group starts with what the founders call "library time," a focused stretch where participants work on their own issue. Berger and Willdorf move around the room, helping people choose a tool, refine a prompt, or adjust their approach when the first answer misses the mark.
This feels more like a working session than a presentation. Nobody needs to memorize a list of AI features. People learn by attempting a task they already understand - then checking whether the tool actually reduces effort or improves the result.
Why the ground rules matter
Each gathering comes with rules of engagement. One is an "ideological parking lot": the event isn't meant for a wide-ranging debate about whether AI will save or ruin society. That doesn't mean risks get ignored. Berger acknowledges that the technology can feel threatening, especially to creative workers. The narrower purpose is to build fluency, test tools, and reduce administrative load.
That boundary is useful because AI conversations often bundle multiple questions into one:
- Is this technology socially beneficial?
- Is a specific tool reliable for this task?
- Is it safe to share this information with the provider?
- Does using it improve the work - or just produce more output?
Those deserve separate answers. A focused test won't settle the big philosophical debate. But a broad debate also won't tell you whether a tool can summarize a document accurately enough to save you time.
The Society's format avoids another common trap: assuming "using AI" automatically means asking a chatbot to write everything. Some participants don't want AI to draft for them. The facilitators instead help them explore other uses - summarizing, organizing, comparing, or drafting a first version that a human can review.
Where AI can create practical value
The examples Berger and Willdorf share tend to involve work people already need to do. They're not made-up demos built around novelty. They're attempts to reduce the time spent navigating complex information or making a professional decision.
One attendee had a question about her mother's eligibility for long-term care under an insurance plan. The insurer sent a 300-page packet. Berger says the attendee used AI to review the material and received an answer indicating her mother qualified - along with a draft letter to the insurer.
That story points to a plausible use case: helping someone locate relevant parts of a large document and plan the next step. It doesn't prove the AI correctly interpreted the policy or that the insurer would agree. The output still needs to be checked against the policy terms, and a high-stakes coverage decision may require advice from a qualified professional. Still, reducing the initial burden of finding and organizing information can be useful by itself.
Administrative work is not automatically low-value work
Administrative tasks can be repetitive and still carry serious consequences. Reviewing a care policy, preparing a client proposal, or sorting through complicated instructions may look "administrative," but mistakes can affect money, access to services, and working relationships.
That's why "automate the boring stuff" isn't a complete strategy. The better question is whether AI can assist with a limited slice of the task - while verification and final decisions stay with the person who understands the stakes.
Berger also described using Claude to review a consulting proposal after she realized she'd priced the work too low. She asked the tool to flag weaknesses and make the proposal more robust. The AI suggested her fee didn't match the scope and recommended starting by doubling her rate. She followed that advice, and the client accepted without objection.
This isn't proof that a chatbot can always set the right market rate. The useful role here was closer to a second opinion: spotting a possible mismatch between scope and price that Berger could then judge and decide on. The final call remained hers.
Economic value is a better test than novelty
The founders emphasize the economic value of saving time and improving access to information. That's a more grounded test than asking whether an AI feature looks impressive in a product demo.
For an individual, value might mean fewer hours building a first draft or understanding a document. For a business, it might mean faster turnaround on a recurring process, fewer handoffs, or more time for work that actually requires expertise and judgment. Either way, the gain needs to be weighed against the cost of checking outputs, protecting data, and keeping the workflow running.
As Berger puts it, AI can work like a new household appliance: a tool that takes some repetitive labor off your hands. The analogy has limits. A washing machine runs a predictable cycle; a language model generates probabilistic responses that can still be wrong. Time saved only counts if the review work doesn't swallow up the minutes the tool was meant to give back.
The limits: fluency is not blind trust
The Society's approach shouldn't be read as "everyone should use AI for everything." The founders say they're not advocates for every aspect of the technology. Their events focus on experimentation, and a good experiment also includes the possibility that a tool is the wrong choice.
AI may be a poor fit when you need a guaranteed correct answer, when the information is highly sensitive, or when the cost of an unnoticed error is simply too high. A polished response isn't proof of accuracy. The tool can omit a condition, misunderstand what you asked, or give a confident answer without enough evidence behind it.
Keep sensitive information out of unapproved tools
Before you upload an insurance document, medical record, client file, or employment information, check the tool's data-handling terms and your organization's policies. A personal account may not offer the same privacy protections as an approved business service. In regulated environments, rules like GDPR - or other sector-specific requirements - may also apply.
If you don't know how a provider handles submitted information, don't assume it's private. You can often test the workflow first using a redacted document, a sample, or non-sensitive information. It's a small operational step that can prevent a much larger problem.
Verify claims against the source
When AI summarizes a contract, policy, or other consequential document, ask it to point to the exact passages supporting its answer. Then read those passages yourself. Treat the output as a navigation aid, not a substitute for reviewing the actual terms or seeking qualified advice.
A simple verification process might include:
- Checking important claims against the original document.
- Asking the tool to identify uncertainty or missing information.
- Comparing the answer with another reliable source when appropriate.
- Having a qualified person review high-stakes decisions.
- Keeping a record of what was AI-generated and what you confirmed independently.
Fluency includes knowing how to use a tool - and when not to trust it. The point isn't to become dependent on a chatbot. It's to improve your ability to decide whether a tool belongs in a workflow at all.
A practical framework for trying AI
You can apply the Society's approach without organizing a dinner party. Start with a real task and a controlled test. The framework below is designed to keep the experiment small enough to evaluate, but serious enough to learn from.
-
Choose a recurring, bounded task
Pick a task that takes real time and has a clear beginning and end. For example: extracting key dates from a document, summarizing meeting notes, comparing two drafts, or drafting a first version of a routine email.
Avoid starting with a vague goal like "use AI to be more productive." If you can't measure anything, you'll never know what improved. And you may end up with extra "activity" rather than better work.
-
Define what a good result looks like
Figure out what would make the output useful before you run the tool. That might mean a summary that captures every deadline, a draft that needs light editing, or a reduction in time spent finding relevant sections.
If you can, set a baseline. How long does the task take now? How often do errors show up? If the AI-assisted version takes almost as long because it requires extensive checking, that's still an important result - it just means the experiment didn't save time.
-
Give the tool context, then iterate
A weak prompt often produces a weak answer. Explain the task, provide relevant context (only what's safe to share), specify the format you need, and make clear what the tool should not assume.
Then iterate when necessary. If the answer misses the point, refine the request rather than treating the first response as "good enough." Just don't spend 20 minutes tweaking prompts to save two minutes on the work. Measure the whole workflow, including setup and review.
-
Review the output and record the trade-off
Check factual claims, names, dates, calculations, and any recommendations. Then compare the finished result with how you handled the task before.
Did the tool save time? Make the work clearer? Make decisions easier? Or did it introduce risks or add extra review steps?
Keep using the tool only if the benefit is clearly worth the cost. That cost includes subscriptions, staff training, data controls, and the ongoing effort of checking outputs.
-
Share what worked - and what did not
One advantage of the Society's mixed-profession gatherings is that participants can compare approaches. A nonprofit worker, a designer, a parent, and an executive may face different problems, but a useful prompting or verification technique can still travel across settings.
For teams, the equivalent is a short internal record of tested workflows: the task, the tool, what data was off-limits, the review steps, and the outcome. That turns one-off experiments into something teams can build on - without pretending one success story means a workflow is instantly scalable.
Conclusion
The Aged Intelligence Society offers a practical model for AI adoption: start with a real problem, learn by testing, and keep judgment with the person who understands the consequences. That approach holds up better than adopting tools just because they're trendy - or rejecting them because the risks sound scary in theory.
AI can help with administrative work that eats time, especially when the task is bounded and the output can be checked. It can't replace expertise, privacy safeguards, or critical thinking. The next step is modest: pick one task you already understand, test an approved tool to see if it genuinely makes things easier, and keep the workflow only if the results justify both the review effort and the risk.
For the founders' original interview and further context, read WIRED's conversation with the Aged Intelligence Society.