Professor, Department of Infor-mation & Communication Sci-ences, Sophia University19algorithms today that I did not see in 1977 when I started study-ing artificial intelligence.”“The recent breakthroughs in AI are due to the huge amount of data now available to drive the technology,” Marwala noted. He explained that in 2000, when he wrote his PhD thesis, a neural network would have around 30 input units. “Today, 10,000 input units is considered a small number.” Marwala introduced several cases of AI contributing to society within the framework of the UN’s Sustainability Development Goals (SDGs). For example, in relation to the third SDG, “Good Health and Well-Being,” Marwala outlined how AI was used to predict the spread of COVID-19 in South Africa, which empow-ered the country to efficiently prepare its medical response. “We predicted the peak month in South Africa well before it ac-tually happened,” he said and explained the application of SIR model parameters – a simple mathematical model of epidemics that charts the spread of diseases – to AI to predict the spread of the pandemic more accurately. “We found that this allowed for much more precise results and had excellent positive implications for health and well-being.” Another example Marwala gave of AI contributing to health and well-being had to do with the development of an artificial larynx that he designed with other University of the Witwatersrand sci-entists in Johannesburg, South Africa. For decades, people who lost their voice due to cancer or other health issues could only hope to communicate in the robotic voice of a mechanical larynx. This new artificial larynx, however, comes close to actual speech by approximating tongue movements. AI and sustainable developmentIn general, AI uses data for three purposes: to predict, to clus-ter, and to generate. A predictive AI might be used to predict the effects of climate change or war, while a generative AI, such as ChatGPT, may be used to produce various types of content, in-cluding text, images, audio, and synthetic data. “From the movement of the tongue, the device can discern what someone is actually saying,” he said. “The movement is translated into a signal, which is then taken to a neural network that inter-prets the person’s actual words.” Marwala cited this device as an example of AI contributing to good health and well-being, in ad-dition to innovation.AI’s advancement is forcing society to confront issues of privacy and surveillance, algorithmic bias and discrimination, and job displace-ment as potential risks of AI’s tech-nological evolution. Regarding the future of AI, Gonsalves said he be-lieves the next major advancements will be in “artificial emotional in-telligence” and “cognitive artificial intelligence.”“Artificial emotional intelligence will be able to simulate human intuition, experience, and creativity, capabilities lacking in today’s ‘narrow’ AI. Cognitive artificial intelligence will simulate human reasoning. Right now, narrow AI is just capturing patterns in data. No reasoning is involved. Future AI will be able to learn from con-text. It will be flexible and adaptive to different situations.”Contemplating AI’s potential, Gonsalves asked, “What if AI goes in the wrong direction?” Dr. Gonsalves on the need for “AI ethics”Following Marwala’s talk, Professor Tad Gonsalves shared his perspec-tive on the future of AI and the need to establish a set of ethical standards for its use. He said, “My belief as an AI researcher and an educator is that we need to use our superior wis-dom to design artificial intelligence ethics. Right now, the AI and algorithms we have are like huge beasts, or dragons, eating and digesting tons of data and giving us predictions. We need to train AI to have ethical values. My model would base the conscience of AI on the human conscience of knowledge, experience, empathy.” Gonsalves believes that we can tame the artificial intelligence dragon by combining external rules and regulations with an internal ethical conscience.“The best thing we can do is guide the develop-ment of AI, rather than try to stop it, because it’s almost unstoppable at this stage.” Tad Gonsalves
元のページ ../index.html#19