Artificial Intelligence is "thinking" for us?

In the last seven days, many of us may have gone to the internet, opened a social media app or specifically shopped online, and, without even trying, we may have used AI in some capacity. That is how pervasive these Large Language Models (LLMs) have become, and with that degree of presence, it is pertinent to question – “How do LLM-based AI models affect us?”

The polarity of the discourse about using A.I.- with one group touting it as an omnipotent being which would replace people in almost all (if not all) fields of work and the other resorting to shaming people who use it- has made it difficult to have nuanced conversations about it. But that makes it all the more important to study the long-term impacts of AI on us as individuals and how we function as a society.

Let’s Understand Critical Thinking

Critical Thinking is our ability to make decisions through understanding and analysing information after properly organizing them. People who think critically can not only evaluate the content of the information they are studying; they can deduce and interpret the context of it too.

For example, A person who thinks critically about fitness-related information on social media would probably not only be able to deduce how legitimate the information is; they could also analyse the effectiveness of the information based on contexts like the target audience they are targeting, country, culture, general availability of food sources, etc.

Needless to say, critical thinking requires our active participation in understanding any piece of information.

So why are we bringing A.I. into critical-thinking conversations?

When most people gained access to generative AI in 2022, they mostly used it to search for information, but within a few months, it became part of people’s lives in more ways than one. Students use chatbots to aid with their school/college work (in extreme cases, students passively copy the content generated by AI chatbots), use it to generate images/videos, and to have “conversations or therapy among other things. The natural language processing gave them a somewhat "human-like" attribute, leading many people to engage with these chatbots as proxies for friends, therapists, teachers and more.

A lot of conversations around A.I. have made researchers worry about how the general public has been using it to understand the effects on our cognition and capacity to think.

A paper written in 2025 highlighted cognitive offloading (the practice of letting someone else-in this case, A.I. Chatbots- understand, analyse and evaluate information on our behalf) and how it relates to our ability to critically think about anything. Predictably, there is a relationship between how often we use A.I. chatbots to think for us and our critical thinking skills. People who used A.I. tools often scored low on critical thinking skills compared to those who didn't. There is a chance that people who might already not practice critical thinking may rely on technological tools for the same. But this paper still highlights the need to understand what could happen with long-term use of A.I. Chatbots.

The conversation is not black and white, because there can be benefits of these tools when we use them to engage in deep conversations, but it can hamper our learning when we use them to structure an answer for us.

Another paper, from 2011 (before A.I. chatbots were mainstream), talks about the "Google Effect", where easier access to information affects our memory and also problem-solving skills. The inferences from this paper can be applied to LLM chatbots, and further research can give more insight into their effects.

Many people may also think of A.I. tools as being a reliable source of information, and people have used these tools for advice. Most popular chatbots use conversational, friendly language that can also make people "feel like" they are talking to someone, which can add to the trust they feel towards these tools.

There have been reports of people trusting information from A.I., and the results were grim. Recently, a farmer in China destroyed 25 acres of his crops after seeking and following the advice from A.I. tools.

 Last year, an article mentioned a 60-year-old man developing bromism (symptoms can range from anxiety, hallucinations, tiredness, muscle loss, poor ability to walk and more) after enquiring with a chatbot about replacing table salt. One argument people may have about these articles/news reports is that the people using A.I. might have been older in age and hence have less technological literacy. But there have been cases where lawyers have used A.I. to form their legal arguments, resulting in false citations of cases being added that don't exist in real life (A.I. Hallucinations), leading to disruptions in legal proceedings.

What do we do now?

With the popularity of LLM chatbots like ChatGPT, Claude, or Gemini, the world seems to have varied opinions, ranging from people hailing it as the next big thing in tech to creative professionals, like artists, writers or musicians, expressing displeasure at the unethical nature of A.I. training, where human creative work is used to train A.I. mostly without the consent of the artists (Meta uses the voices, artwork or real images shared on Instagram or Facebook to train A.I. models). This article is not about the ethics (or lack thereof) of A.I. or the impact (negative effects on the environment, human reviewers being hired who have to review people's questions in the chatbots or being exposed to harmful images or content during the review process) it can have. People who are vehemently against A.I. use will not be able to stop the majority of people from using these tools, and shaming others is one of the least effective ways to create awareness. But we can start with small steps:

A.I. literacy and Critical Thinking: Introducing classes on critical thinking and A.I. literacy is the need of the hour. Knowing what A.I. is and how it affects us is essential for people. Generative AI, when used to do the "thinking" for us, can negatively affect our ability to analyse information and be patient about the process of generating ideas. Not only that, training AI can have long-term impacts on the environment (the effects of this are already seen in places where households are experiencing astronomically high electricity bills, poor quality of water, among other things), and human reviewers (who review the information put in by other people in the chatbots or other AI-related platforms) have also been subjected to extremely sensitive content (that has left many traumatised) during the AI training processes while getting little financial compensation and very poor safety measures. Knowing how human lives are affected due to AI use is essential in knowing what the future could look like for many of us.

Directing towards less harmful alternatives: Dr Fatima, an educator, talks about an intervention method used in addiction intervention called Harm Reduction. This is a strategy. It was originally used to help people who abuse substances do so in a way that is less harmful while they were receiving help to make the transition easier for them (Eg., hospitals providing fresh needles to people instead of using old/used needles, thus reducing the vulnerability to suffering from HIV/AIDS). Using Harm reduction methods to regulate AI use, especially among people who cannot (required by jobs) stop using these tools, to ensure the least possible harm to our environment, ethical use of information and ensuring our data stays safe.

Using LLM/AI chatbots to engage as an additional tool: While we cannot decide for others whether people will use these tools or not, we could create awareness of the possible long-term effects of passive use of AI tools. Using them for deeper analysis as an additional tool to cross-check them, rather than making these tools "think" for us, could be a better way to engage with these tools.

Understanding the Ethics of using AI: Generative AI doesn't take an ethical route in producing its content, and there is naturally a sense of frustration and anger among creatives, researchers and the general public alike. From finding shady ways to use work created by actual people to train AI (which uses up massive amounts of water), to people using chatbots that encourage sycophancy when people use them as "therapists" or friends- these are very few ways these companies have already caused harm to people and the intellectual property of others. 

As long as major companies keep on investing in LLM chatbots, this technology will continue to pervade every part of our lives, and we might not even be given a choice. In such situations, shifting focus to harm reduction rather than glorifying or shaming the use of AI might have a better long-term impact that ensures more people make informed choices.