Το AI μπορεί να κάνει τα πάντα, εκτός από το να πάρει την ευθύνη | Νίκος Μακρής | TEDxEleusis

AI Technology Evolution - AI algorithms are advancing rapidly, doubling in capability in under 6 months, significantly faster than the 18 months for computer transistors based on Moore's Law [6][7] - The breakthrough of large language models, stemming from research papers in 2017-2018, has led to AI becoming a general-purpose technology impacting various aspects of life [4][11][12] - The field is evolving towards AI producing new knowledge, requiring algorithms to advance further to generate insights from patterns across different scientific domains [18][19] AI Applications and Impact - AI is impacting education through personalized teaching, with over 70% of educators viewing it positively [13] - AI is transforming various sectors, including healthcare, with AI-developed drugs entering clinical trials as early as 2020 and breakthroughs in protein structure prediction in 2023 [14][15] - AI enables creative content generation, including presentations, music, and art, exemplified by AI tools like Ghost Writer [16][17] Challenges and Considerations - AI systems are data-dependent and can perpetuate biases present in the training data, highlighting the need for awareness regarding uploaded content [23][24][25] - AI training and inference are energy-intensive, requiring GPUs that consume approximately 10 times more power than CPUs, leading to significant energy demands [26][27] - The industry needs to establish guard rails and ethical frameworks for AI development, as reflected in the EU AI Act, to ensure human oversight and address potential risks [35][36] - Deepfakes pose a significant threat due to their realistic nature, requiring heightened awareness and caution when consuming online content [32][33][34]