Does AI Think?

The question of whether AI can think has been a topic of debate for quite some time. Renowned computer scientist E.J. Dijkstra famously remarked that asking if AI can think is akin to asking if submarines can swim. His point, perhaps, was that as long as AI accomplishes its intended tasks, why should we concern ourselves with whether it can truly "think"? However, in our society, we debate over whether a hot dog is a sandwich or if cereal qualifies as soup. It seems only fitting to explore the concept of AI thinking as well.

In today's AI race, the ability to claim that an AI model can think and reason holds significant value. It's not just philosophical musing; it's about tangible advancements and market appeal. For instance, OpenAI's latest models have been touted as "thinking and reasoning" models, a claim designed to excite stakeholders and boost stock prices. While these models are undoubtedly impressive, do they genuinely "think"?

The dictionary defines "think" as "to form or have in one's mind." Simplified, computers lack minds, and neural networks do not really count. Some may argue this definition is too narrow or outdated. So, let's consider a more commonly understood definition.

According to Merriam-Webster, "thinking" is defined as:

1. The action of using one's mind to produce thoughts, or the process of considering or reasoning about something.
2. Opinion, judgment, or a way of thinking about something, such as a particular outlook or belief system.

These definitions highlight both the cognitive process and the conclusions or opinions that result from it.

By Merriam-Webster's definition of "thinking" as the action of using one's mind to produce thoughts or engage in reasoning, AI does not technically "think" in the same way humans do. Here’s why:

1. AI lacks a mind: While AI systems process information, they do not have a conscious mind, self-awareness, or subjective experiences. AI operates based on pre-programmed algorithms, statistical models, and data patterns, not the intentional or reflective mental activity humans engage in.
2. AI simulates reasoning: AI can perform complex tasks like pattern recognition, decision-making, and problem-solving, which appear similar to human reasoning. However, this "reasoning" is based on programmed instructions and large amounts of data rather than true understanding or internal thought processes.

In essence, AI can simulate aspects of thinking, but it doesn't "think" in the sense of having its own opinions, judgments, or independent reasoning like a human mind does.

Personally, I believe thinking implies a level of intentionality and perhaps even consciousness—benchmarks that AI, in my view, has not met thus far.

Perhaps Dijkstra was right all along—we shouldn't be having this conversation.

What are your thoughts? Can AI truly think, or are we merely anthropomorphizing our sophisticated tools?

Kim Brian

Modern Applications and Power Platform Solutions Architect at Velrada .

Technical Consultant Helping organizations unlock the full potential of their Microsoft efficiency tools.

Feel free to share your thoughts or connect with me to discuss AI or Microsoft efficiencies.