The Five Barriers to Artificial Intelligence. Braunschweig, B. Futuribles, April, 2023. Place: Paris Publisher: Futuribles
The Five Barriers to Artificial Intelligence [link]Paper  abstract   bibtex   
The launch in late November 2022 of ChatGPT, a conversational agent (chatbot) developed by OpenAI, received a great deal of attention and has prompted a lot of reporting and commentary on the performance of, and advances made by, artificial intelligence (AI). And yet, though progress in this area is indeed impressive, a number of limits remain and the emergence of a totally autonomous, perfect AI still lies in the realms of science-fiction. For example, Bertrand Braunschweig, in charge of the scientific coordination of the French programme 'Confiance.ai', points here to five major 'barriers' that could well hinder its development if insufficient attention is paid to them. He writes of these here: they have to do with trust in AI, its energy consumption, the safety of the systems controlled by AI, human-machine interactions and, lastly, the inhumanity of machines. In closing, he suggests a number of lines of work and research for meeting the challenges posed by the five barriers: improving network architecture, combining digital and symbolic models, increased interdisciplinarity and other measures.
@article{braunschweig_five_2023,
	title = {The {Five} {Barriers} to {Artificial} {Intelligence}},
	issn = {0337-307X},
	url = {https://www.proquest.com/scholarly-journals/five-barriers-artificial-intelligence/docview/2783968819/se-2?accountid=14542},
	abstract = {The launch in late November 2022 of ChatGPT, a conversational agent (chatbot) developed by OpenAI, received a great deal of attention and has prompted a lot of reporting and commentary on the performance of, and advances made by, artificial intelligence (AI). And yet, though progress in this area is indeed impressive, a number of limits remain and the emergence of a totally autonomous, perfect AI still lies in the realms of science-fiction. For example, Bertrand Braunschweig, in charge of the scientific coordination of the French programme 'Confiance.ai', points here to five major 'barriers' that could well hinder its development if insufficient attention is paid to them. He writes of these here: they have to do with trust in AI, its energy consumption, the safety of the systems controlled by AI, human-machine interactions and, lastly, the inhumanity of machines. In closing, he suggests a number of lines of work and research for meeting the challenges posed by the five barriers: improving network architecture, combining digital and symbolic models, increased interdisciplinarity and other measures.},
	language = {English},
	number = {453},
	journal = {Futuribles},
	author = {Braunschweig, Bertrand},
	month = apr,
	year = {2023},
	note = {Place: Paris
Publisher: Futuribles},
	keywords = {Artificial intelligence, Attention, Barriers, Safety, Coordination, Energy consumption, Fiction, Human-computer interaction, Social Sciences: Comprehensive Works},
	pages = {1},
}

Downloads: 0