Can AI really think?
AI is everywhere – revolutionising how we interact with technology, reshaping industries and transforming businesses. But behind the all the advances, there’s a question: can AI really think?
If you’re a coach, consultant or small business owner looking to understand AI’s true capabilities and limitations, read on…
More than a simpler spellchecker
Some might dismiss AI as a glorified spellchecker or predictive text tool on steroids – “it merely predicts the next likely word to form a coherent sentence.” To some extent, this is correct. AI’s basic principle rests on identifying what comes next in a sequence. However, AI’s speed and scale of prediction puts it far in advance of a simple spellchecker, making it incredibly powerful. Instead of linear, rule-based computing, an AI tool utilises a vast network of neural-like connections. The result is language generation that often feels surprisingly human. It’s this that gives the impression that the machine is thinking.
High-level mimicry
As far back as the 1840s, the farsighted Ada Lovelace wrote what is often viewed as the first software, and algorithm for a never-built in her lifetime the ‘analytical engine’. Lovelace predicted that such machines would eventually move beyond mathematics to carry out creative tasks, like playing music. However, she also foresaw that while such machines would excel at following instructions, they would lack the ability to think or organise data independently. Over a century later, computing pioneer Alan Turing would dub this limitation ‘Lady Lovelace’s Objection’.
Turing is probably most well known for devising what is now known as the Turing Test to address the potential question of the ‘thinking machine’.
Turing decided that ‘thinking’ was too difficult a concept to define and called his test the “imitation game” – the game being whether a machine could imitate human responses sufficiently well to be mistaken for a person. If it could do so, it could be considered ‘intelligent’.
Today’s AI models often pass the Turing Test with ease, leading people to feel as though they’re interacting with something that thinks. However, for all the impressive responses, AI is designed to predict rather than understand. The human-like language and nuanced text simply reflects the level of technology itself, not the presence of true thought.
Today’s AI may appear sophisticated, but Lady Lovelace’s Objection still stands – AI can make predictions, generate text, and even simulate creativity, but it lacks true understanding or independent thought.
AI doesn’t need to ‘think’ to change the world
A lack of thinking ability doesn’t stop AI from being transformative. Just like computers in general, AI’s explosion in popularity is down to a simple economic principle: as a technology becomes cheaper, it is used more widely.
Widely accessible, cheap large language models (LLMs) and transformer-based AI have made processing massive amounts of data more efficient and affordable.
And as the cost of using AI decreases, its applications expand. While the hype surrounding AI is undeniable, so are is the impact: cheaper prediction technology is already changing our world, and its influence will only grow.
How prediction can transform your business
Maybe AI can’t think but its imitation of intelligence is still a game changer. For example, predictive AI models can:
- manage stock inventory by predicting demand patterns,
- optimise logistics by predicting the best routes,
- control automated production, relying on robots.
And then, of course, there’s the use of AI models to create marketing materials (as described in many other articles on this blog).
As AI prediction becomes cheaper, businesses that rely on human-led prediction will face competition from those leveraging AI for the same tasks. This creates both opportunities and challenges for small businesses. The key is identifying where prediction can add value to your operations and where it can disrupt and overtake traditional methods.
The best prediction machine we’ve built (so far)
ChatGPT and other large language models put this new level of predictive capability into the hands of business owners. No, ChatGPT and other LLMs are not ‘thinking’ as such, but by processing billions of pieces of data, these systems are capable of generating human-like responses to complex prompts.
Already, AI is being used to predict credit card fraud, automate loan applications, and has ‘watched’ enough videos of humans driving to accurately simulate human driving behaviour. This isn’t thought; it’s refined data processing on a monumental scale and it has significant implications for fields as diverse as finance, customer service, and content creation.
So, back to the question: Can AI really think?
By now, it should be clear that no, AI cannot think (at least, not yet). Whatever the hype says, AI is merely(!) an advanced, data-driven tool designed to predict outcomes based on patterns, not true understanding or cognition. That said, AI is still reshaping industries in profound ways, creating opportunities that are economically possible for businesses of all sizes.
The next stage of AI development is making decisions, and that work is well underway with some models already able to do this complex task. How reliable they are and how ready people are ready to trust these systems remains to be seen; after all, would you trust an AI that can’t count the number of R’s in word Strawberry to make a decision for you? There’s still a way to go until we get to this point.
The challenge is to look beyond AI’s imitation of intelligence and see it for what it is: a powerful prediction tool whose strengths you can harness for your business – delivering value and staying ahead of the competition.
