Three years of ChatGPT and a world split between AI believers and skeptics
Three years after ChatGPT launched, the world has split into two pretty distinct camps: people who use AI tools in their daily lives and people who still prefer to keep their distance. And this divide is growing faster than you might think.
A Pew Research Center survey, published in 2025, revealed that one-third of American adults now use ChatGPT on a regular basis. Among young people under 30, that number climbs to 58% — practically double what it was two years ago.
With all this growth, experts are warning that it is time for a more open conversation about how to use these tools in a way that actually makes sense — and with responsibility. 🤖 After all, just using them is not enough. You need to know when, how, and most importantly, what for.
With that in mind, we talked to experts to understand the best approaches for taking advantage of AI in areas like creativity, research, learning, and organization — without giving up human judgment in the process.
What experts say about responsible AI use
The conversation around responsible use of AI tools goes way beyond simply avoiding mistakes or not sharing sensitive data. According to experts in technology and digital behavior, the central point is understanding that AI works best when it complements human reasoning, not when it tries to replace it entirely. That means before handing off any task to a tool like ChatGPT, Claude, or Perplexity, it is worth getting clear on what you want to achieve with that response, because the quality of what you get back depends directly on the quality of what you ask.
Another point that keeps coming up in responsibility discussions is verification. AI tools can generate incorrect information with impressive confidence — what experts call hallucination — and this is a real risk, especially when the topic involves data, dates, names, or any information that needs to be accurate. Catherine Goetze, a content creator and AI educator known as @askcatgpt on TikTok, drives this point home: check your sources, check the links, and check the dates on those sources. The consensus among researchers in the field is that users need to maintain the habit of verifying everything, especially when the generated content will be shared publicly or used in professional and academic settings.
There is also the question of long-term impact on skill development. Teachers and educators have been raising an important warning: when people delegate cognitive tasks too frequently to AI — like summarizing texts, building arguments, or solving problems — they can end up losing the practice of doing those things on their own. Goetze herself illustrates this trap well: she uses ChatGPT to help her come up with content ideas, but if she asked the tool to write scripts that she just read word for word, she would be using AI as a crutch that undermines her creative ability. This does not mean that using AI is bad, but rather that there is a balance to be found.
Brainstorming: AI as an idea partner
If you are not sure where to start, experts are unanimous: use AI to generate ideas. Timothy B. Lee, author of the Understanding AI newsletter, recommends using AI tools for brainstorming and breaking tasks or projects into actionable steps. According to him, whenever you are trying to come up with ideas, AI is a good starting point.
Goetze takes it a step further and suggests thinking of the tool as a thinking partner. The idea is to use it for bouncing around ideas, pushing past creative blocks, and refining your reasoning. But there is a key detail here: when analyzing the results, you need to keep trusting your own judgment, your experience, and your personal taste, rather than letting AI have the final say. The best tasks for AI, according to her, are the ones where you already know what the right answer should look like. 🎯
This approach takes the pressure off anyone who thinks they need to master some fancy technique before they can start using the tool. In practice, it is a conversation where you present a problem and AI helps explore possibilities you might not have considered on your own.
AI as an ally for research and learning
One of the most promising applications of AI tools is right in the area of research. For more complex or research-intensive projects, an AI tool can give you an overview of what has already been published on a given topic. Lee draws a pretty useful comparison: think of AI similarly to Wikipedia. We know it is fallible and we know how to check the citations.
Tools like Claude, ChatGPT, and Perplexity offer some variation of a feature called deep research. This functionality scans documents on your topic and summarizes them into a report of a few thousand words. The bot even asks clarifying questions to put together the most relevant results for what you need.
Lee describes the feature as incredibly good. The responses include primary sources and links, allowing you to check the original materials yourself. According to him, the deep research function is excellent for getting a lay of the land — identifying key articles to read, questions to answer, and next steps for your project.
The essential point, on which all experts agree, is that after using an AI tool for this preliminary guidance, you still need to do the real work on your own. AI speeds up the initial mapping, but the deep dive, the critical analysis, and the conclusions remain the responsibility of the person conducting the research.
The role of AI in research is also changing the way people learn new subjects. Instead of relying on a single book or course to understand a complex topic, you can now use an AI tool to get a general overview, identify the points that deserve deeper exploration, and create a personalized learning path. This is especially useful for professionals who need to stay constantly updated in fields like technology, medicine, law, and finance, where the volume of new information is massive and the time available to consume it all is always less than what is needed.
Learning new skills and hobbies with AI
You can also use AI when it allows you to expand your world, says Ella Hafermalz, an associate professor at Vrije Universiteit Amsterdam who studies the impact of generative AI on work.
Hafermalz has been using AI to get help growing a mini lotus flower, to learn the basics of stock investing, and even to get dinner ideas. According to her, the benefit is in getting you from zero with a new interest, activity, hobby, or skill, especially those with a high barrier to entry.
For a lot of people, the obstacle to trying something new is embarrassment, fear, lack of time, discomfort, or simply not knowing where to start. Messaging back and forth with AI can break down those barriers and get you to the next step, as Hafermalz explains. But once again, it is best to treat it as a starting point for lower-risk tasks, where you remain the final authority. 🌱
Organizing information: where AI truly shines
If there is one area where AI tools show undeniable value, it is organization. Once you have started your research project, AI can help structure your findings — identifying themes, answering questions, or generating timelines and summaries.
If research is your priority, Hafermalz recommends using Google NotebookLM, a free app that works only with documents, notes, and materials you upload yourself, rather than pulling information from across the internet. Historians are already using NotebookLM as a research assistant, she says. The tool works as an organizer: if you are trying to organize information and synthesize content without wanting the AI to go off-script and pull random stuff from Reddit, NotebookLM offers a more controlled space.
For personal use, AI can work like a household or executive assistant, helping you plan meals or workouts, create a budget, or organize your afternoon by prioritizing tasks from high to low priority.
Tools integrated into platforms like Notion, Microsoft 365, and Google Workspace already allow AI to summarize meetings, categorize documents, prioritize tasks, and even suggest schedules based on your priorities. For teams working remotely or freelancers juggling multiple projects at once, this kind of support can save a significant amount of time and mental energy — freeing up space for what really matters at work.
How to get better results from AI
About a year ago, crafting the perfect prompt was considered essential for getting the best out of an AI tool, but that is becoming less and less important over time, according to Lee. The models are evolving and responding more intuitively, even to more casual and informal language — though providing context still helps a lot. The more information you give, the more likely you are to get a good result, he adds.
Goetze encourages people to ditch the idea of prompting as a rigid technique altogether. In her view, the best approach is to think of it as a conversation. The magic actually comes from the back and forth, the exchange of messages. That is the advantage ChatGPT has over Google search, she adds: go into the tool and just say what is on your mind.
As long as you are careful not to share sensitive information, you can also send website links, PDFs, and other materials so the AI factors them into its response. For example, you could share your new phone contract and ask the AI to flag important clauses or potential savings.
The reverse prompt technique
If you hit a wall at any point, there is a technique called the reverse prompt. When Goetze recently had a mental block on a document she was working on, she asked ChatGPT to come up with five questions that would help her push past the obstacle. The questions led her to reflect in slightly different ways, she says. It is a creative approach that flips the traditional logic of using the tool and can unstick processes that feel stalled.
The risks of relying too much on AI
It is essential to verify every response generated by AI. While they are getting better every day, they still need to be checked against primary and reliable sources, as experts agree. The model can not only repeat false information and misinformation — it can actively make things up, which is known as hallucination.
It is also possible to use AI in ways that shrink our world, undermine our abilities, and even our humanity. Often, these problems come up when we treat the tool as a shortcut or as the final step of a process, instead of a first step.
All three experts we consulted agree that it is going to become increasingly difficult to tell which parts — or what proportion — of a text were written by humans and which were generated by AI. For now, Goetze recommends being transparent about your use and being careful to avoid plagiarism and copyright violations.
Another potential danger is investing too much in AI responses and even becoming dependent on them. Hafermalz is pretty direct on this point: you do not want to get stuck in a feedback loop with AI — you will end up in dark places.
Her recommendation is to set a clear goal or intention every time you use ChatGPT, and gradually increase the complexity of your use to make sure you always stay in control. AI should not be a prison that keeps you trapped. It should be a stepping stone — a path to getting out and doing other things, as Hafermalz puts it. Use it where you can verify things for yourself, in the real world.
How to take your first steps mindfully
For anyone still just starting to explore AI tools, the best advice experts have been giving is to start with real, everyday problems. Instead of testing the tool with generic questions, try identifying a specific task that eats up unnecessary time or energy — whether it is organizing meeting notes, drafting an email, summarizing a long document, or researching a topic for a presentation — and see how AI can help with that particular thing. This practical, focused approach tends to produce more concrete results than aimless exploration.
Throughout the process, it is crucial to maintain a critical eye toward what is generated. That does not mean distrusting everything, but rather building the habit of reviewing, questioning, and adapting AI responses before putting them to use. The more you interact with a tool, the more you learn to ask better questions — and better questions lead to more useful and accurate results. This learning cycle is what transforms a beginner into someone who actually knows how to tap into the potential of these technologies.
Finally, it is worth remembering that the AI tools landscape is constantly evolving. New versions, new models, and new features pop up at a pace that can be hard to keep up with. Staying at least somewhat up to date on what is available — without the pressure of testing everything at once — is a smart way to make sure you are using the best that technology has to offer, in a conscious, safe way that lines up with your goals. 🚀
