Instead of divine inspiration from the Bavarian Aloisius, AI now has to do the job!
AI is supposed to make everything so easy. You type something in and the solution comes out.
Basically like the famous and much-discussed divine inspiration.
But is it really all that easy? Simply being able to sing “Hosiana” is not enough.
Besides, the question is whether AI even understands the singing.
But jokes aside. AI seems to be the solution to all of humanity’s problems. Well, at least for those a software user has.
In any case, that is what the software vendor seems to want you to believe.
“We use AI in our solution” is one of the most common statements in software marketing, and many people fall for it.
Whether AI is a practical implementation here, whether enough data is available, or whether employees in front of the screen can actually do anything with it is a completely different matter.
But that does not seem to bother the many AI believers. Because only a vendor that supposedly uses AI is a “modern” vendor that also delivers added value. Not a chance.
The problem is often quite simple. Software users are far too quick to be dazzled by the claim “AI included” without even knowing what AI really is and how its use in their own company can deliver added value.
Simply buying an “AI module” and then thinking the world will become a better place is a widespread misconception.
With this blog post, I would like to try to shed a bit of “light on the AI fog” . If you are looking for even more enlightenment, feel free to get in touch and we will put the much-praised AI functions of software vendors to the test during the purchasing process.
As always, it will be exciting. Be sure to keep reading! Because waiting for the “divine inspiration” like the servant Aloisius from Ludwig Thoma (A Munich Man in Heaven) is not an option!
By the way: with separate AI projects, the topic can be viewed somewhat differently in parts. Today, I am focusing on “AI included” in operationally used solutions.
Continue in the blog post.
Aloisius is still waiting for divine inspiration while others think they have already found it!
(drawn by ChatGPT)
What is the difference between an optimization algorithm and artificial intelligence?
Old wine in new bottles?
In the past, people always talked about an optimization algorithm. Today it is supposedly AI. But what is the difference, actually?
The difference between an optimization algorithm and artificial intelligence (AI) mainly lies in the application area, the goals, and the complexity of the respective methods:
Optimization algorithm
An optimization algorithm is a method for finding the best solution to a given problem.
- Goal: maximizing or minimizing an objective function (e.g., cost, time, efficiency).
- Application: often used in mathematics, logistics, engineering, and business.
- How it works:
- Has a fixed set of rules to find the best solution
- Often relates to problems with defined constraints and a clear structure
- Examples:
- Linear optimization (Simplex algorithm)
- Evolutionary algorithms (e.g., genetic algorithms)
- Gradient-based methods such as gradient descent
Artificial intelligence (AI)
Artificial intelligence is a broader concept and includes systems or algorithms that perform tasks that normally require human intelligence.
- Goal: problem-solving, decision-making, pattern recognition, and often learning from data.
- Application: used in areas such as image recognition, natural language processing, robotics, and games.
- How it works:
- Can it use optimization algorithms to achieve its goals?
- Often data-driven: AI models can improve by learning from experience (e.g., machine learning)
- Examples:
- Neural networks (deep learning)
- Decision trees and random forests
- Reinforcement learning (e.g., AlphaGo)
What is the key difference?
- Optimization algorithms:
- Are tools with clearly defined rules
- Have a fixed goal and no ability to learn or adapt to new data
- Artificial intelligence:
- Can include optimization algorithms, but is much more general
- Has the ability to learn from data and adapt to changing conditions
An example to illustrate the difference:
- An optimization algorithm could calculate how a delivery service plans the shortest routes
- An AI could additionally predict traffic volume, adapt to new patterns, and even suggest future routes based on historical data
Is there real AI behind this, or just a “simple” algorithm?
From the Stone Age to AI heaven
Take a close look at the image below. Then take a moment and ask yourself:
- How modern are our processes already?
- How much AI knowledge is already available in-house?
- How modern are the technologies and tools we use?
- How well integrated are all our systems?
- How good and clean is our data, and is it available in sufficient quantity?
If you then, for example, realize that your “tool” in IT currently is probably still equivalent to a Stone Age club, you should ask yourself whether you are even ready to set off on the journey to AI heaven.
Yes, there are certainly isolated solutions where this is possible almost regardless of status. But the “big leap”, for example in transport management, will only succeed if the rest also fits.
Moving from a Stone Age organization with the corresponding tools to AI heaven is a real challenge
The “prompt”—the answer to all questions!
But learning how to ask questions and create instructions is essential!
One version of AI that can be used very well in presentations is to ask a question/define requirements in a “prompt” and the AI immediately spits out the answer. Of course, all based on your own data.
But with “analyze ….”, “optimize ….” or “show me options for ….” it is not that simple. Even if the demo quickly shows great answers or even charts on the screen, you still have to ask how practical these results really are.
- Was all data in the system really taken into account?
- Which data was taken into account, which was not, and why?
- Did I ask the question correctly?
- Is there even enough data in the system to get a meaningful answer?
- What happens if I ask the question again? Or if someone else asks it and deviates only slightly in the wording? Which answer is the right one then?
As the saying goes, roughly: most of the time, the clueless person is sitting in front of the screen! Simply accepting the displayed answer as “God-given” (sticking with our theme) would be a serious mistake. Especially if you want to make a business decision on that basis, it could end fatally. The “heaven” you end up in may be quite different from what you planned.
The assumption (see the error of assumption in my book “Digitalization in Industry, Retail and Logistics“)
In logistics, simply replacing a dispatcher with AI and a “prompt” is, in my view, an outrageous claim. The “reverse gear” is pre-programmed.
The input prompt. Example: ChatGPT
Always be the first to receive the latest news, interviews, and expert articles?
Nothing works without data
One thing should really be clear to everyone. Even “artificial intelligence” needs data. And lots of it. It does not work with your own brain either, does it?
AI-based optimizations, for example in operational systems, have a serious problem. Anyone switching to new software, for example in transport management, is missing something essential. Exactly: your own data!
On what basis is AI supposed to analyze? Without a “mountain” of data and the use of human intelligence (depending on how individualized the application is), nothing will happen at first.
By the way: Also ask whether your data is being transferred to a “central data mountain”. A possible issue especially with cloud solutions. Why should others learn from my data? (This could also have something to do with data protection or even disclosure of trade secrets!)
The “data mountain”
(drawn by ChatGPT)
Put AI sales promises to the test: We can help you
Anyone who wants to know whether the AI praised so highly in the software solution really delivers value or is practical—and not just looks great in the sales process—should seek support.
For a free initial consultation, simply request an appointment here via the link. It is definitely worth it!
Using AI still requires some brainpower when selecting it!
Spot fake AI: 3 simple criteria
You can tell, for example, by these 3 criteria whether what it says on the tin is actually inside:
- Marketing buzzwords: Is AI used only as a buzzword without a concrete technical implementation?
- Rule-based systems: Is it just a pre-programmed rule-based system with no ability to learn?
- No updates: Can the software not be trained on new data or updated?
Conclusion: Anyone who does not know what AI actually is, what application areas exist in the company, or how to truly test practicality has a problem. But there are solutions!
Image sources: ChatGPT, Shutterstock
