 Artificial Intelligence
| 'Summarize the most important events of 1968 for me!' |
No, we would never ask AI that, nor would we even speak to AI in that tone. We would proceed much more cautiously. We would be significantly less assertive because we want to 'keep control' of the conversation.
Coincidentally, yesterday we learned from a published interview with a teacher how many students misuse AI. And the question above is just a harmless example. They use AI to save work, take the first three facts and
that's it.
This has similarities with private leasing, if you'll pardon the comparison. You don't really acquire something, you have to pull yourself together considerably afterwards, and in the end you are still judged by others.
If they are very picky, there is trouble in the air.
| The students' behavior is probably even worse than private leasing.
|
Of course, this is something different here. But we would always ask questions in such a way that we could rule out mistakes as far as possible, buying in such a way that we could not be treated unfairly. Of course, this is
something different here.
It's quite simple when it comes to computer problems. This practically goes without saying, culminating in the question: Does one of the solutions suggested by ChatGPT work or not? Because there is one thing
AI cannot do: fully comprehend the respective configuration.
Things are always different than expected. For example, the tip is actually good for Windows 10, but useless for Windows 11. You can't find the directory where you're supposed to do something. AI does not always fit
perfectly.
That is the first thing to criticize about the question above. It should be very well thought out. Don't just ask questions at random. Limit the space for play and thinking. If possible, test the AI by asking questions that you know
the answer to.
Our questions, whether technical or general in nature, are usually much longer. Their quality can be recognized in part by the long consideration periods. We often break down what we want to know and approach the
actual topic a little more cautiously.
We also ask more politely, not in this commanding tone. If ChatGPT helped us, then we would like to express our gratitude. We don't know if it has any influence, but that's just how we are. However, that doesn't mean
that we are slowly beginning to confuse this AI with a human conversation partner.
We are particularly careful with technical data. Only the devil knows where she got this data from. Always double-check, force yourself to draw on other sources, and then cross-check using simple mathematics.
Even worse would be the following example, which is very similar to the one above: Who invented the first direct gasoline injection system? Was it Rudolf Diesel who immortalized something like this in a sketch, or was it
Moto Guzzi, but only in a test engine?
Were the Americans perhaps earlier, Mercedes in a 601 aircraft engine or Gutbrod with a two-cylinder two-stroke engine? Is the sketch considered a 'first draft', 'trial', or even 'series'? You may notice how important it is to
ask questions as precisely as possible.
| 
