POVZETEK - ChatGPT je v zadnjih nekaj letih povsem spremenil pomen besede "umetna inteligenca", kot smo jo poznali do sedaj. Milijoni dnevno rešujejo svoje težave s pomočjo te tehnologije, ki se vedno bolj zažira v vse pore našega življenja. Toda tako kot vsaka stvar ima tudi ChatGPT svojo ceno. Ali ste vedeli, da ChatGPT dnevno porabi kar 621.4 MWh električne energije? Prevedeno na CO2 emisije to znaša približno 275 ton dnevno! Toda ali obstaja alternativa? V zadnjih letih se pospešeno razvijajo tehnologije na področju optičnega računanja, ki bi lahko pomembno pohitrile kompleksne računske operacije, ki jih lahko najdemo v modelih umetne inteligence kot je ChatGPT. Lahko ta tehnologija vodi do optičnih nevronskih mrež, ki bodo izboljšale učinkovitost modelov in zmanjšale njihovo porabo električne energije?

ABSTRACT - In the last few years, ChatGPT has completely changed the meaning of the word "artificial intelligence" as we know it. Millions of people solve their problems every day with the help of this technology, which is increasingly embedded in every pore of our lives. But like everything, ChatGPT also has its price. Did you know that ChatGPT consumes 621.4 MWh of electricity per day? Translated into CO2 emissions, this amounts to approximately 275 tons per day! But is there an alternative? In recent years, technologies in the field of optical computing have been rapidly developing, which could significantly speed up the complex computational operations that can be found in artificial intelligence models such as ChatGPT. Could this technology lead to optical neural networks that will improve the efficiency of the models and reduce their electricity consumption?