AI as the ideal employee or "just" a useful tool?
This article appeared in the automation special of Primeur, the trade magazine for the AGF sector.
The most outrageous marketing claims are flying at us from all sides: “AI is going to take over our jobs,” a fear shared by many people. The reality? Largely exaggerated statements that, to this day, have delivered few concrete results. AI has turned out to be an umbrella term that is not always used honestly in practice. Think of companies that claim to be “AI-driven” while only using simple automation or statistical analyses. The suggestion is created that AI is an all-encompassing technology that can take over “everything,” while a correct application of the technology requires a high degree of precision from (yes indeed) humans. How do we, as a software supplier, view AI and its current applications?
So what is AI, really?
Artificial Intelligence is the technology that enables systems to exhibit intelligent behavior by analyzing their environment and, with a certain degree of autonomy, taking action to achieve goals. Quite a mouthful. Let’s break this down a bit:
An AI system operates based on algorithms and data. These two components combined determine the quality of the system. If the algorithm is incorrect, the AI system won’t work. If the data is incorrect, the outcome will be incorrect as well. You therefore always need to be certain that both components are correct; only then can the AI system produce a meaningful calculation.
Augmented Intelligence as the most common application
AI is currently most often used in combination with human involvement. Humans design the system’s control and evaluate and process the output. This is therefore referred to as “Augmented Intelligence” rather than “Artificial Intelligence.”
A good example of Augmented Intelligence is McDonald’s Christmas commercial. Last Christmas, McDonald’s Netherlands launched an alternative AI-generated Christmas ad that caused a great deal of controversy in the creative industry. So much so that McDonald’s removed the commercial everywhere not long afterward. The head of the creative agency behind the commercial (The Sweetshop) said in an official statement (which has since been taken offline) that over a period of seven weeks, no fewer than ten employees worked full-time on the AI production and “barely slept” during that time. It likely cost The Sweetshop more time and labor to refine the AI commercial than it would have taken to create it in the traditional way. And all this while the criticism of McDonald’s mainly boiled down to a “cheap and lazy approach.” Here too, you can see that the perception of AI and its actual application are completely out of sync.
Should we be afraid for our jobs?
Martijn van Andel, director of software company jem-id, puts the claims into perspective: “AI systems work based on pattern recognition, enabling extremely fast calculations with the right prompts. Especially for exact, binary problems, AI is unquestionably better than humans. So the potential is enormous. Think of it as a smart, fast employee who never says ‘no,’ has perfect grammar, and clearly explains how an answer was reached. However, it also regularly gets things completely wrong, while the explanation sounds flawless. And that’s exactly where it gets tricky. AI does not think logically on its own, but retrieves information from sources that are made available to it, such as the internet or a database. That seems very reliable, but not everything on the internet is correct. The technology is not yet advanced enough to act fully autonomously and question its own conclusions. For that, it is simply not reliable enough yet.”
Many market players cleverly tap into entrepreneurs’ fear of ‘missing the boat.’ Striking examples are used to paint an ideal picture, while false promises are made. Such as recognizing handwritten orders, while in reality no one ever receives handwritten orders. Technically very impressive, but practically nobody is needing it.
Conclusion: how do you distinguish value from false promises?
AI is a powerful technology that, when used correctly, can be of great value to companies. It can speed up tasks or provide new insights when facing bottlenecks. However, we see companies investing large sums in AI due to half-truths and false promises, often because they lack the right knowledge. Most people simply have limited technical expertise. That’s not a criticism, but it does put certain claims into perspective. When the conversation focuses mainly on the technology rather than the business problem being solved, that is a clear warning sign. Although many AI applications have proven to be effective, the term is often used incorrectly to present a utopia. To avoid disappointment and high costs, we want to send a clear signal with this article. Logical thinking and human expertise remain crucial for the correct application of AI. There are many other great technologies that, when applied in the right context, can lead to major efficiency gains. AI is simply one of them.
How jem-id uses AI
For GreenCommerce, the ‘Greenius AI assistant’ was developed using GPT and Azure OpenAI, Microsoft’s cloud service that makes the power of ChatGPT securely available. The assistant supports GreenCommerce by providing support staff and developers with faster answers in code and documentation, and by helping new employees during onboarding. In addition, Copilot supports the development teams in software development and fine-tuning systems.
Using Retrieval-Augmented Generation (RAG), a technique in which an AI first retrieves relevant documents and then uses that information to formulate a reliable answer, Greenius searches only within controlled, internal data. This keeps the answers reliable and secure. Because Azure OpenAI does not use data for training purposes, sensitive information remains fully protected.
Would you like to know more about our use of AI or discuss possibilities? Then contact us at info@greencommerce.nl or call +31 174 642 622.
