Is AI as bad as it’s been made out to be… or even as good as it’s been made out to be?

AI has not left public conversation in years. AI is bad, AI is profit, AI will replace us all, AI is the future, AI is killing the planet. 

Arguments against AI defend human intelligence, pushing for integrity behind art and the human touch. Arguments promoting AI claim convenience and technological revolution, marketing it as highly profitable. AI is integrated into every app and search engine, yet it seems that people are spending more time managing AI tools than they actually save by using them. 

Attitudes are very divided: the industry side is optimistic and investing heavily, while public sentiment is one of AI fatigue. So, after the last few years of supposed “innovation,” what is there to show for it? What are we mistaking as advancement, and what happens when the smoke clears? 

Wait, What is AI?

Science fiction novels and Black Mirror episodes often forewarn of AI gaining sentience and replacing humans. Tech companies also play along with this idea of rapid development of a tool that can replace humans in the workforce, selling this promise of innovation to investors.

It is important to dissect these narratives and understand what AI actually is. AI is not an overnight invention growing beyond control; it has been used in machine learning models in the past for many professional and scientific fields. It enables systems to learn patterns from data, gradually improving performance by training itself. Autocorrect on a Word Document uses these traditional machine learning models, placing that squiggly red line under a misspelled word or suggesting the next word based on the user’s most frequent typing patterns.

Under the umbrella of machine learning are large language models (LLMs), used for chatbots like ChatGPT and Gemini. LLMs are designed to recognize patterns in text and generate human-like responses. They don’t think, feel or possess awareness. What is often mistaken for “intelligence” is just prediction and pattern recognition. 

When given an input, LLMs calculate the most likely next word based on training data. There’s no inner experience, intention or understanding behind it. The myth of intelligence and expertise persists because LLMs are very good at mimicking conversation, creating the illusion of personality or consciousness. However, this is less like a human mind and more like autocomplete on a keyboard. 

The AI Boom and Bubble

When Wall Street told tales of exponential profit in the stock market, it crashed and led to the Great Depression. When widespread confidence in the U.S. housing market proved to be wrong, it led to the 2008 housing crisis. NFTs and cryptocurrency saw the same hype and sharp fall, leaving investors out to dry. 

Economic bubbles are a market phenomenon that describes rapid increases in asset prices,  where the promised value of an asset does not match the asset’s actual intrinsic value. It usually plays out like so: inflated prices caused by optimism in market behaviour and speculative investment, causing an eventual burst of these bubbles when market sentiment catches up and firms and investors bail and incite rapid sell-off. 

Many economists noticed a similar trend surrounding the AI industry. The AI bubble hypothesizes that the massive hype and investment surrounding the AI industry exceeds actual profit, similar to previous economic bubbles. Companies are now participating in an AI arms race of sorts, investing heavily in AI by building data centers, launching startups and racing to adopt new tools, all driven by fear of missing out. 

Despite this investment, firms like OpenAI are burning cash. In 2025 Open AI collected $4.3 billion in revenue while still posting a net loss of $13.5 billion during that six month period. OpenAI is not the only one. An economic report from MIT concluded that despite investing US$30-$40 billion into GenAI, 95 per cent of organizations are not seeing returns on their investments. The same companies are cycling the same investments amongst themselves, creating an illusion of profit and abundance within the industry. 

This is where cyclical investment patterns come in that explain the illusion of endless growth in the AI sector. In the early phase of a tech boom, companies overspend on infrastructure like data centers, chips and capacity, anticipating explosive future demand. At the beginning, this can look like growth and a promising industry. That’s exactly what firms have begun to do in this AI boom, stockpiling to avoid falling behind in AI. If everyone is spending, there must be a good reason, right? 

The risk is that this turns into overinvestment. If demand doesn’t keep pace or if companies realize they’ve built more capacity than they can profitably use, spending slows sharply. This has been observed in other tech bubbles, where there is intense buildup, followed by a cooling-off period when reality catches up to expectations. 

The AI boom has created a bottleneck that runs straight through the AI chip maker NVIDIA. Most modern AI systems depend on powerful graphics processing units to train and run models. Whether it’s startups or Big Tech firms like Microsoft, they’re all buying the same core ingredient: NVIDIA chips. The entire AI industry is stacked on top of a single supplier, causing their profits to surge. In this AI digital gold rush, many are saying that NVIDIA is selling the shovels. 

With the sale of its AI chips, NVIDIA’s total profits for the fiscal year was USD$120 billion in 2025, which has grown hugely since the $4.4 billion of profit that they saw in 2023. This means that the entire tech industry is investing in physical infrastructure and supporting the growth of one firm, which also happens to be the only firm to see any profit in this AI boom. 

Public Perception

Over the last few years, AI has developed and improved at a rate that was hard to foresee, with regulations lagging behind. In the early days of OpenAI and DeepSeek, the public met generative AI with curiosity but skepticism. Many were immediately against a tool so inaccurate and dysregulated. Others were excited by the consumer-level access to services like ChatGPT and, seemingly overnight, being able to generate texts and images at the drop of a hat. Though intrigued by its ability to mimic human speech, generative AI engines did not yet have the access to information needed to maintain accuracy, nor flawlessly mimic human speech. 

Today, AI is an educator’s worst nightmare; teachers rush to reel in AI usage among students while AI continues to improve and better mimic human speech but not necessarily factual accuracy. Government regulation has also not moved as fast as the development of AI, which has led to society working around it rather than effectively integrating it. According to a 2025 study by Abacus Data, 51 per cent of Canadians distrust AI, and 55 per cent reportedly never or rarely engage with it. 

Despite the narrative that AI is a magic wand that the public and professional sectors will integrate, there is major distrust and skepticism stemming from a lack of regulation and exploitative data gathering. According to the study, the top five concerns of the Canadian public regarding AI are malicious uses of AI, spread of misinformation and fake content, loss of privacy, safety and security concerns, as well as increased unemployment. 

On the other hand, the leading perceived benefits of AI are increased efficiency and productivity, reduced human error, enhanced convenience, improved healthcare and medical advances and better access to information and education. The study found that 45% of Canadians use AI tools on a regular basis. This means that for some, AI is a neutral tool that has become a part of everyday life. This is true for younger Canadians especially: the study found that 72% of those aged 18 to 29 and 62% of those aged 30 to 44 regularly use AI tools in their everyday lives.

AI’s Achilles Heel 

Though there is distrust, public perception agrees that the potential is there. An article by  

The Kuwait College of Science and Technology published in 2025 looks at how students and faculty apply AI tools in post secondary education. 

Titled “Integrating artificial intelligence in higher education: perceptions, challenges, and strategies for academic innovation”, the study found that both students and faculty recognize AI’s potential in enhancing teaching and learning, while simultaneously distrusting its reliability and ethics. The article argues that while AI tools have value in settings like the classroom, barriers to AI integration exist because of a lack of ethical guidelines and training. 

Clearly, AI is undeniably powerful but currently highly capitalized; privatization and subsequent smoke and mirrors surrounding AI is impeding its value in making meaningful changes in the way it was promised. 

The good news is that the technology is still very real. The artificial intelligence models are undergoing real advancements that have undeniable gains. This means that without market forces and lack of regulation, AI can live up to the expectations of providing net benefit in sectors like research and healthcare. 

While other economic bubbles and fads burst as fast as they ballooned, AI’s trajectory may more closely resemble the dot-com bubble during the 2000s. The dot-com bubble occurred when the tech industry theorized that the internet would change the world. Though those predictions panned out to be correct in the long haul, the timeline and returns promised to investors were highly inflated at the time, causing stock markets to crash and a majority of tech startups to go under. However, after the bubble burst, the technology and infrastructure was what remained, and the world relies on the internet to this day. 

AI has the potential to be everything it was promised to us, without the ethical concerns, the lack of integrity and the dysregulation and that it has thus far turned out to be. The first step is to understand where these inflated narratives are stemming from, so we might understand not only the real harms but also real advantages. With proper government regulations and ethical guidelines on both usage and development of AI, the industry might begin to course correct in the right direction. By addressing the barriers that challenge AI has all the potential to add, rather than subtract from humanity.