AI AGAIN, AGAIN AND AGAIN…
The hype around AI has been going on for months. I read somewhere that everyone has become an AI expert lately, and how true that is. It writes the tests for the students, the questions for the teacher, advises on court rulings, draws the Pope in puffy jackets, brings the end of the world, and cracks our passwords… And now I’m giving you another shot at a heart attack…
If we have access to such technology, where are the APTs (Advanced Persistent Threat, government-backed hacker groups), governments and the various specialised services? By the time we, as ordinary citizens, see the results of these developments, by the time the ‘tinfoil moon landing’ is a tinfoil moon landing, will the intelligence services working under a few acronyms be dead? But don’t worry, it doesn’t stop our hearts.
Let’s also finally acknowledge that AI is not yet Him. Although the huge mass of information base and the proper programming of language procedures make the answers we get look like those of an entity with a consciousness and personality, it has neither consciousness nor self-image and its creativity is a statistical processing unit with a good random number generator.
A few days ago there was a round in the media about AI wanting to destroy humans (ChaosGPT), but if we set aside the clickbait title, the completely lay journalistic spin, we see something else. If you define the destruction of humanity as the goal as the input of a program and set the algorithm in that direction, then even a google search for the most powerful weapon of mass destruction (spoiler: Tsar bomb) is only a clever toddler’s level of a program called evil, if that is an incredible technological feat. In fact, if ChaosGPT were the unchallenged master of nuclear bombs after such an instruction, we’d be in trouble. But for now, let’s stick to the text game level.
Another cry from the internet ether was that AI can decode our passwords in seconds. However, the table uploaded alongside the article only reveals what was true years ago – that short, easy-to-guess passwords are not worth a damn, as a modern password cracking tool can decode complex passwords of less than 12 characters in minutes. Again, it is not clearly stated in the article that this does not require AI. So it is not because of that, but because of the general evolution of IT tools that multi-factor authentication, password passwords and caution are required when using passwords. The risk is, however, that an AI-enabled system, using existing tools, will be able to exploit vulnerabilities at what speed, since we can already optimize even our malware code with AI.
While we have the finite application of predicting depression based on text reading, AI will not necessarily talk to me on a medical level if I turn to it with a complex problem of my mind and emotions. It’s not the machine that makes me suicidal, it’s the inhuman and often emotionless society. And the developer should not be responsible. If I cut my wrists, the blade would not be to blame.
The whole point of this whole thing is that misinformation, sound bites and titles steal the show. Instead of thinking, looking for and learning the right ways to use AI, we just absorb the posts that appear in the columns of gossip magazines and on the sleazy social media.
I believe that it will be a question of responsibility or failure. If I know who is responsible for what activity, if I know that social norms are respected when using a tool, in this case AI, if I know the context in which it works, the possibilities – but that is true for the internet as well – then it will be like a bread knife… I could stab the neighbour with it, but I usually just eat a good slice of buttered bread.
Sources:
https://www.hwsw.hu/hirek/66012/homesecurity-mi-jelszo-feltores-passgan-jelentes.html
https://hvg.hu/tudomany/20230414_chaosgpt_mesterseges_intelligencia_alapulo_chatbot_gonosz_tervei
http://www.ma.hu/eletmod.hu/387648/Ongyilkos_lett_egy_apa_miutan_egy_AI_chatbot_rabeszelte
Ask our experts!
