Artificial Intelligence has emerged as a topic of great interest and concern, generating intense dialogue across different media platforms. Its ethical, economic, and social implications continue to be debated, and its potential applications within companies have become a subject of much discussion. As businesses explore AI’s capabilities, they must also consider the challenges and opportunities it presents, including its impact on society.

IT professionals are eagerly embracing the opportunity to lead AI adoption in their business operations. The potential benefits, whether real or perceived, are a welcome addition to the institutional narrative, further fueling their enthusiasm. Though not statistically significant, a quick look at LinkedIn can provide an idea of this trend.

It’s essential to underscore that AI is not just another technology to add to an organization’s arsenal. Its adoption necessitates a pragmatic approach, with clear goals, detailed strategic planning, and a comprehensive understanding of its potential and limitations.

Furthermore, assessing the long-term commitment required to develop AI-based solutions and the real economic sustainability of related investments over time is necessary. Lastly, the most relevant issue is that the integration of AI into companies, like any technology, should be driven by goals aimed at solving concrete problems. Many IT managers overlook this trivial premise, and the implications that derive from the obsessive desire for innovation at any cost turn sustained investments into sunk costs.


Beyond the Hype

AI is not just a passing trend, but a transformative force that will inevitably impact us all. Unlike other technologies that may have faded, artificial intelligence is here to stay, reshaping the world as we know it. In recent years, only blockchain has generated enthusiasm comparable to AI. However, the most optimistic forecasts have often proven exaggerated, and many blockchain startups are still striving to scale and maintain their market impact.

Although there are significant use cases in this area (such as the De Beers Group), the costs and complexities involved have led to widespread, if not disappointing, results on a large scale. On this topic, I invite you to read the article The Truth about Blockchain, published in the Harvard Business Review in January 2017, for further insights. There has indeed been recent discussion about the potential of blockchain in ensuring the security and accuracy of AI algorithms. However, its actual applicability and convenience still need to be demonstrated.

In the context of AI, especially in service companies, some IT managers consider it only as a tool to increase business productivity, neglecting its critical applications such as data management, trend forecasting, process optimization, and creating new business models. AI-based virtual assistants have become the obvious first solution proposed by everyone. However, limiting the discussion on AI to this level is reductive, primarily because such tools represent only some of the answers to the problems they promise to solve.

Take ChatGPT, for example, an advanced virtual assistant that demonstrates the potential of AI in programming and natural language processing. Despite its remarkable progress, ChatGPT is still subject to significant errors. It is essential to understand that these systems could be better and that the user must still have knowledge of the topics they seek assistance with.

For example, I’ve asked ChatGPT to solve a simple irrational equation:

ChatGPT: Sure! Here’s a concise explanation of the solution:

To solve \(\sqrt{x+3} = 1 + \sqrt{2-x}\):

  • Isolate one of the square roots: \(\sqrt{x+3} – 1 = \sqrt{2-x}\).
  • Square both sides: \(x + 4 – 2\sqrt{x+3} = 4 – 2\sqrt{2-x} – x\).
  • Move all terms with square roots to one side: \(2\sqrt{x+3} – 2\sqrt{2-x} = 0\).
  • Simplify by dividing both sides by 2: \(\sqrt{x+3} – \sqrt{2-x} = 0\).
  • Square both sides again: \(x + 3 = 2 – x\).
  • Solve for \(x\): \(x = -\frac{1}{2}\).
  • Substituting \(x = -\frac{1}{2}\) into the original equation confirms its validity.

Therefore, the only real solution is: \[x = -\frac{1}{2}\]

This solution is wrong. I won’t go into the calculations’ details, but the equation’s correct result is (x = 1).

When I pointed this out, the ChatGPT responded as follows:

My apologies for the oversight. Let’s correct. The only real solution is \(x = 1\).

WolframAlpha’s solver, on the other hand, was more accurate, correctly answering the initial query.

Despite the uncertainties, generative text tools like ChatGPT are formidable for many tasks, such as coding in various languages. The returned code is almost always functional, if not optimal. However, it provides a starting point, drastically reducing development time and the frustration of developers facing doubts or new problems to solve. However, these are palliative solutions and do not represent a disruptive AI-based technology capable of profoundly altering the organizational fabric of companies.


Conclusion

It is more than a certainty that AI is here to stay and will profoundly impact all industrial sectors. However, it is equally plausible to believe that many of its applications in companies, especially in these initial hype stages, will quickly be forgotten.

  • I recommend reading The Age of AI by Henry KissingerEric Schmidt, and Daniel Huttenlocher, September 2021. It offers a clear and forward-thinking vision of AI’s future scenarios.

  • The Age of AI has begun, by Bill Gates, March 2023.