Artificial intelligence has become the biggest tech trend of the last decade, showing incredible results in the right applications.

It’s also a technology rife with overpromising, where marketers often speak as if future potential has already been fulfilled. This can leave anyone working in supply chain tech in a difficult position, wondering if AI should be used or ignored.

The truth is somewhere in the middle. While we might not know it, more basic forms of AI tech already exist in the sales tech every business uses. More advanced generative systems, while important to watch, can exacerbate existing problems if not implemented carefully.

The Simple Examples are Best

The most basic definition of artificial intelligence is any tech that simulates human brain activities like learning, decision-making, and autonomy. By this definition, any business that uses any degree of software automation and sorting is already using artificial intelligence. We don’t call this AI, just like we don’t call smartphones computers, it’s just more business as usual.

A popular example in sales is perfectly demonstrated by modern real estate systems. For example, consider if a user went online to ask “How can I value my house?”. The answer from this service is a database system used to connect buyers to sellers. With free cash offers and no hidden fees, these websites still use advanced computer sorting systems, but not modern AI. The results in this case are reliable tools and trusted output.

Problems with the ‘Cutting Edge’

There are two main problems with the current implementation of cutting-edge generation AI solutions. The first is that a lack of oversight and transparency regarding their work processes makes them difficult to track. Modern generative AI platforms can create enormous portions of modern apps, which might seem to work initially.

Problems arise if a business doesn’t employ people who understand the code, so it’s impossible to ensure that the system uses proper practices instead of shortcuts. You won’t know if the errors are due to human error or an AI mistake, so eliminating bugs can be difficult and costly.

The second issue with generative AI comes from its capacity to interpret and implement detail where none exists. This is commonly called hallucination, and it’s the result of a machine feigning confidence and authority despite having no actual understanding of data and context. Even when the AI software is understood, the data the AI pulls can be interpreted incorrectly, leading to dangerous mistakes.

An example could come from when an AI is asked to pool information from a client into one document. AI could sort through thousands of data entries instantly, but inconsistencies and errors can be misidentified and overlooked. This could result in client data being duplicated or overwritten or vital information being removed altogether.

The main takeaway from all this is that while advanced AI is promising, it’s still nowhere near fulfilling this promise reliably. Instead, it’s the basic and traditional forms of AI that are proven and trusted, having already powered successful business systems for decades. There may be a time when generative systems can be trusted in business applications, but that time has not yet arrived, so it can be best to let others test the waters.