Without hints: how to use AI and not forget how to think on your own
Neural networks help you write texts, prepare presentations, and analyze complex topics, but transferring the entire task to them can interfere with the development of your own skills. ""Izvestia" was told on September 25 how to use AI for learning, to test their independence and to negotiate with children about working with new tools, a methodologist and product specialist of corporate training, founder of the agency for methodology WES (We Educate the Smartest) and the School of the Product "Method", vice-president of the Association of Entrepreneurs of online Education Natalia Kundera.
First your own attempt, then the neural network
According to the expert, when mastering new technologies, people often make the same mistake: first they perceive the tool as a way to speed up work, and then they discover that, along with routine operations, they have transferred part of their own thinking to it. Therefore, before contacting the neural network, it is necessary to determine which skill a person wants to train.
If the task is to learn how to write, Kundera recommends that you first prepare your own text and only then ask the AI to find weaknesses. When preparing for an exam, it is useful to answer a question yourself, and when solving a problem, take at least the first step without prompting.
"If you need to learn how to count, count for yourself first. An erroneous self—response is often more useful for learning than an ideal neural network response, because it provides material for analysis," she explained.
The four roles of AI in learning
Kundera identified four roles in which a neural network can help learning: examiner, opponent, interlocutor, and explainer. In the first case, instead of asking the AI to outline the topic, it is worth asking the AI to ask five questions, not to suggest answers and after each of them to point out omissions. This approach is suitable for exam preparation, job interview, or presentation.
As an opponent, the neural network can look for weaknesses in the user's position and challenge it. According to the expert, it helps to train argumentation. To prepare for a difficult conversation, AI can be asked to play a client, supervisor, or colleague, ask uncomfortable questions, and disagree automatically.
Another technique is to ask a neural network to explain a topic first to a child, then to a teenager, and then to a specialist. The next step a person must take on their own: retell what they have learned in their own words. If this does not work out, the material has not yet been mastered, Kundera noted.
Separately, the expert advised not to instruct the AI to immediately correct the errors found. For example, instead of asking you to make a letter to a client more professional, you can ask the neural network to specify three places that may raise questions and explain the reasons. After that, the user makes the changes himself.
"It's worth asking the AI to find the bug first, not fix it. When a person discovers a problem by himself and redoes the result, he gets a skill, not a ready answer," the specialist emphasized.
This approach, she says, is applicable to resumes, presentations, homework, program code, and exam answers. In corporate training, mastering a skill is also associated with a sequence of actions: a person sets a goal, tries to solve a real problem, receives feedback and makes a new attempt. A neural network can speed up this process if it remains part of the practice.
At the same time, Kundera warned against building education around an unchanging list of "skills of the future." She cited data from the World Economic Forum, according to which by 2030, about 39% of existing employee skills will change or become obsolete. In these circumstances, the expert recommends developing the ability to retrain: ask questions, check information, notice mistakes, explain your own decisions, and receive feedback.
Independence check: a task without hints
To assess how much a person relies on a neural network, Kundera suggested trying to complete a familiar task without it.
"If it works, but it's faster with a neural network, the tool helps you. If you don't know where to start without it, some of the skill may have been gradually transferred to the machine," she said.
In such a situation, the expert advises to periodically set up a "no-prompting mode": write a text yourself, solve several examples without a calculator, read an article without automatic retelling, or formulate your own position before contacting AI. According to her, this practice allows you to train independence.
How to negotiate with a child about using AI
Kundera recommends that parents negotiate with their children about the use of neural networks. It involves your own attempt to solve the problem, then turning to AI and finally explaining the result in your own words.
"It is not necessary to forbid children to use neural networks. It is much more productive to agree on three rules: first, your own attempt, then the AI, and at the end, an explanation in your own words," the expert said.
Once a week, you can discuss with your child what he did on his own, what tasks he handed over to the neural network, and how he checked its response. Such a conversation, according to Kundera, helps to gradually learn how to manage technology.
The specialist emphasized that AI can be tasked with finding options, structuring information, asking questions, and identifying weaknesses. However, if the goal of the task is to learn how to think, write, count, analyze, or make decisions, it is important for a person to maintain their own participation in the entire process.
Переведено сервисом «Яндекс Переводчик»
