ChatGPT. In recent weeks, everyone was talking about the chat robot see can generate texts by itself. Unique texts, difficult to distinguish from 'real' ones. It is an example of artificial intelligence, also called artificial intelligence (AI). AI raises not only technical but also moral questions: how will it affect our economy? Can AI help solve problems in our society, and how? But also: what is the impact of AI on human autonomy? This is what several humanities scholars are researching.
One of them is Professor of Philosophy Jan Broersen (1). With a background in mathematics, logic and computer science, he explores philosophical questions raised by AI. For example, how logic-based methods of modeling can help develop forms of responsible artificial intelligence that are controllable and in the service of humans.
Broersen sees that along with almost unbridled optimism about the possibilities, there is also much fear of artificial intelligence. There is much to nuance, he believes. In Radio1 program The Night of, Broersen explains both the opportunities AI offers and the limits to it. And he reassures listeners who fear evil world domination by artificial intelligence.
Professor Jan Broersen talks further about all the possibilities, but also impossibilities, of AI in his Studium Generale lecture "How Stupid is Artificial Intelligence?" (2), on Wednesday, Feb. 22, 2023.
LISTEN HERE TO THE NIGHT OF... 'THE TWO SIDES OF AI' (JAN. 24, 2023) (3)
Professor of Language, Communication and Computing Antal van den Bosch (4) also works on applications of AI in language and communication sciences. In NRC, he discusses the applications of Large Language Models (LLM): computer programs such as ChatGPT that can analyze, summarize, recognize and, moreover, generate large amounts of text.
Van den Bosch is enthusiastic about translation with speech recognition, both for written and unwritten languages. But he also sees downsides to algorithms that deal with language. One problem is, they can reproduce stereotyping and thus perpetuate bias. Van den Bosch talks about this with Jort Kelder in the Radio 1 program Dr. Kelder en co.
In Het Financieele Dagblad, Van den Bosch is questioned about how ChatGPT is used by students: as a teacher, how can you recognize whether a student's work is authentic or written by a chatbot? There is separate software available for this at ChatGPT, but it is not error-free. "One complication, is that the detector sometimes also brands original work by a student as AI: a 'false positive' result. Exam boards need to consider how students are protected from that."
'CHATTING WITH CHATGPT', DR. BASEMENT AND CO (JAN. 28, 2023) (5)
'ARTIFICIAL INTELLIGENCE MAKES OUR EDUCATION OBSOLETE', FINANCE DAILY (DECEMBER 22, 2022) (6)
'IF THE COMPUTER GETS BETTER WITH LANGUAGE THAN WE DO', NRC (DECEMBER 4, 2022) (7)
Resisting the hype of chatbots
There are critical voices, too. University Professor of Translation Science and Translation Gys-Walt van Egdom (8), like his colleagues, sees that we cannot avoid chatbots like ChatGPT, but thinks we might be able to sabotage them. In an opinion piece in the Volkskrant, he writes:
"I believe we can defuse the hype by massively feeding it. Data-driven technology like ChatGPT actually requires paradoxical engagement. I would advise everyone, especially critics, to play endlessly with the tool and commit data sabotage while playing. Give the chatbot what it asks for. Feed it data, but nonsensical data."
'DEFEND CHATGPT BY DISRUPTING ITS ALGORITHMS,' DE VOLKSKRANT (JAN. 18, 2023) (9)
Finally, a reading tip from professor Antal van den Bosch. All the ifs and buts, from working conditions at tech companies to energy-guzzling servers, are discussed in this Dutch-language article on Knack.be. Van den Bosch: "Nowhere is a curve cut. I agree with what all the experts say, observe and speculate."
'THE DANGERS OF CHATGPT: 'WE RISK BECOMING COLLECTIVELY DUMBER', KNACK.BE (JAN. 23, 2023) (10)
https://www.uu.nl/medewerkers/jmbroersen
https://www.sg.uu.nl/sprekers/jan-broersen
https://www.nporadio1.nl/podcasts/de-nacht-van/84054/de-twee-kanten-van-ai
https://www.uu.nl/medewerkers/APJvandenBosch
https://www.nrc.nl/nieuws/2022/12/04/als-de-computer-beter-wordt-met-taal-dan-wij-a4150269
https://fd.nl/samenleving/1461748/artificial-intelligence-maakt-ons-onderwijs-achterhaald
https://www.nrc.nl/nieuws/2022/12/04/als-de-computer-beter-wordt-met-taal-dan-wij-a4150269
uu.co.uk/staff/GMWvanEgdom
volkskrant.nl/columns-opinie/opinion-protect-chatgpt-by-denying-its-algorithms~b3372de1/
knack.be/news/technology/dangers-of-chatgpt-we-threaten-collectively-dumber/
