Generative AI in industry

Must-Have or Show-Stopper?

Few technologies have accompanied us over the past year with as much hype and promise as Generative AI (Artificial Intelligence). And it has also polarized immensely. While some dismiss the technical innovations as mere gadgets, others jump on the bandwagon and want to solve “everything” with AI immediately.

Both extremes are, of course, dangerous.

Generative AI refers to a class of AI algorithms capable of creating new data resembling the data they were trained on. Imagine giving a computer examples of images or texts, and it learns to develop similar ones. This technology is often used for creative purposes such as generating art, music, or even human-like texts. Generative AI can also be employed in image editing, medicine, and other fields to generate data useful for analysis and decision-making. Ultimately, it aims to empower computers to be creative in a human-like way and generate new content.

OVERVIEW

Don't miss out on progress!

On the one hand, we shouldn’t underestimate new technology. Yes, there are still some issues and “growing pains,” but the conclusion that “My business doesn’t need to change!” with the reasoning, “I tried ChatGPT and it can’t do XY; we still need humans for that!” is unfortunately too short-sighted.

Technology will continue to evolve rapidly, and the fact that AI currently can’t do something is no guarantee that it will never be able to. Missing out on progress has always been risky for business; numerous examples illustrate this.

Do not utilize AI according to the "one size fits all" principle!

On the other hand, it is also fatal to unquestioningly want to use a new technology for “everything.” There are meaningful use cases, and there are less meaningful ones. There are also security risks, as our colleagues at sequire technology made very clear to us last year in their research on “Indirect Prompt Injection”. Not only are leaks of sensitive data possible but there’s also the potential for malicious attackers to take over entire chat systems without the user noticing anything. This isn’t a “show-stopper,” but it’s something to keep in mind.

So, how do we deal with it? Our recommendation:

At this year’s DIGITAL COMMERCE SUMMIT automotive & industry CEO Kevin Dewi referred to an example from the area of product configuration:
For example, when configuring a valve block, there are many redundant or predictable steps that an intelligent assistant could take over. Currently, we are showcasing how a natural language request for constructiong a valve block generates a machine-readable JSON document. This already feels very comfortable and convenient for the user in its current state. But we can also go a step further. What if, for instance, a configuration proposal was automatically and directly generated from the customer’s specifications? Then, only minor adjustments by a human expert would be necessary. This would save a tremendous amount of time and increase efficiency.
christoph-endres

CHRISTOPH ENDRES
managing director
sequire technology

This might also interest you