Can AI already be used for drones?

Drones developed by AI image generators

Right now, many questions are being asked about artificial intelligence (AI). The topic is hyped and is worth a headline in the media every day. We have been dealing with AI since the late 1980s.

Right now, many questions are being asked about artificial intelligence (AI). The topic is hyped and is worth a headline in the media every day. We have been dealing with AI since the late 1980s. First we asked fundamental questions about the semantics and a possible architecture. Then, in the early 1990s, we started thinking about possible applications on LISP machines, which were then built specifically for AI development. There were different approaches in the early phase. On the one hand, for example, the basic principles of an AI were described with dialectical laws (Hegel and Engels), other approaches are based on mathematical and cybernetic models. The decisive question, however, was the possible self-generated knowledge of a machine-based AI system. Put simply, we asked: Can a machine learn?

This question is still hotly debated today. It starts with the question: What does learning mean? Currently, AI systems are based on data collections. This can be texts and links of contexts. But image information can also represent a data collection.

Today's AI systems are rule-based. These systems are mostly based on "if-then relationships" or pattern comparisons. An AI system can carry out tasks and classify the result as successful or unsuccessful – again according to appropriate rules. Good AI systems can specify and adapt their own rules according to the results. However, this is not yet learning, since the "knowledge" component is missing. According to "Duden", knowledge is defined as follows: "Knowledge is insight gained through mental processing of impressions and experiences". AI systems cannot generate "insight". Because this component is (still) missing, we also refer to it as a "simple AI". Cross-links to simulations based on mathematical models are not entirely wrong.

This is how an AI image generator envisions the future of drones

But what is the biggest problem with currently existing simple AI? This simple AI makes mistakes, critical mistakes.

What is happening in the "AI market"?

After Microsoft invested in OpenAI (ChatGTP) years ago and now integrated this as a service into its own search engine BING, the alarm bells went off at Google. Google reacted hastily. With the integration of the AI engine into its search engine, Microsoft completely called into question the previous approach to research on the Internet. The basis of Google's entire business model is thus shaken. Only a few hours after the Microsoft presentation, Google presented its own system. The result was disastrous: the system didn't work; it just wasn't ready yet. The presentation – then with screenshots – also showed blatant factual errors in the search results. Google lost USD 100 billion in market value within a very short time. Investors and stock speculators who fell for the marketing hype about AI realized that AI is still in the early stages of development. As FelixBa (Youtube) discovered, the Microsoft AI also contains errors. However, compared to Google, Microsoft was not penalized on the stock market for this reason.

A self-experiment

We then conducted a self-experiment and asked various AI systems about the future of medical logistics with drones. This is an area in which we are familiar. We can evaluate the results professionally. First, we challenged AI image generators (Midjourney, Dall-e 2) to develop a drone for future medical logistics. As engineers, we had some concerns about the proposed results in terms of their flight characteristics, to say the least. But it's still interesting, and for us it's also funny.

If we then feed an AI-powered text generator (ChatGPT) the same question, the result is a fairly general, but not fundamentally wrong text:

Medical logistics with drones has made significant progress in recent years and will continue to develop in the future. Here are some possible developments:

  • Improved Technologies: Drone technologies are becoming more advanced and reliable, which will lead to faster and safer deliveries.
  • Broader application areas: Drones are increasingly being used in new application areas, such as disaster relief, disease surveillance or the delivery of medical supplies in remote areas.
  • Regulation: Regulatory bodies are expected to adjust their regulations on the use of drones in medical logistics to ensure deliveries can be made safely and efficiently.
  • Data Analysis: The use of data analysis will increase to improve the efficiency of medical logistics and to understand how to better tailor them to patient needs.
  • Integration with other technologies: In the future, drones can be integrated with other technologies such as artificial intelligence and the Internet of Things (IoT) to enable even more efficient medical logistics.

Overall, medical logistics using drones will continue to grow in importance and play an important role in providing fast and reliable medical supplies.


(ChatGTP)

A short review

The answer of the AI text generator is not wrong from a technological point of view. However, important elements in the assessment of the future viability of the application are not formulated, or only formulated very imprecisely.

  • Regulations are an important issue. The main focus is on licensing regulations and airspace regulation. The topic could be covered with the 3rd point of the AI text generator answer. This statement would still be acceptable for a newspaper article, but it is definitely not sufficient as a basis for a strategic decision in a company.
  • Data analysis is not a new element associated with drones. We do this with our software every day for our ground transports. The topic will not be treated differently in quality compared to ground transport.
  • The question of profitability is not asked. But this is the crucial question. Can drone-based logistics keep up with the competitor "car" or even undercut it in terms of costs? So far, drone projects have been subsidized or are still operating at high deficits, as far as figures on profitability have been published at all.

How will AI develop and what challenges need to be considered?

The drone has 3 arms for propellers and will certainly not be able to fly like this

UIn order to describe possible developments, it is worth taking a look at current AI-generated projects. The "Seinfeld" project is particularly interesting here. On Twitch "Watchmeforever" an AI-generated sequel to this iconic "show about nothing," will be streamed in a never-ending episode 24/7. Now this never-ending story has been temporarily halted by Twitch. If you analyze the background, you can see that a Twitch AI recognized a homophobic joke in the AI-generated story that is forbidden on Twitch. The series, which actually never ends, was then banned for 14 days by the Twitch AI. If you put the joke in the story in context, at least half of the viewers wouldn't recognize a homophobic joke. However, the Twitch AI does not recognize the context of the story and then blocks the stream. An immature AI controls an immature AI - that sounds like horror, but it's already reality here.

AI is currently being massively hyped in the military sector. AI is used as a sales argument, especially to move the decimal places in the prices of military technology a few places to the right. Especially with drones, this AI is now being praised as "victorious progress". As described above, the AI systems with their data collections used as a basis are very flawed. AI systems in agriculture can already help to save resources with the targeted use of water and fertilizers. In the military field, however, human lives are at stake, in agriculture they are not. A plant that gets too much water due to an error or inaccuracy of the AI can handle it. If an uninvolved person loses his life due to an inaccurate AI, that is unbearable. Armed drones in particular cause more damage than they are really useful. (https://www.heise.de/meinung/Bilanz-nach-20-years-us-drones-attacks-es-meets-meist-the-false-6212380.html). Drones and AI is a dangerous mix, because the use of an immature technology de facto endangers the lives of uninvolved people.

Conclusion

The topic of AI is exciting, and you can use it to generate some funny pictures and interesting texts. But we would like to doubt whether this is already sufficient for practical use, not only because of the examples shown here. The developments with AI have potential, even if it certainly takes a few years before these AI systems will reach an acceptable level of accuracy and relevance. AI can help with decisions, but you should never rely on AI-generated information. In some cases, such as agriculture, positive effects can already be achieved with AI. In other areas, such as military applications or self-driving cars, there is still a need for an expert who can classify and evaluate this information.

Ähnliche Beiträge

German Copters

Die German Copters DLS GmbH ist spezialisiert auf Drohnen-basierende Services rund um Foto- und Videoaufnahmen, ingenieurtechnische Dienstleitungen und die Durchführung von Transporten mit Drohnen.

Kontakt

info@germancopters.de
+49 351 275 411 79
R.Micknass
SkyLoop
Flughafenstr. 59
70629 Stuttgart

Location

German Copters Location