Developing techniques for ensuring AI alignment with human values and goals Investigating potential risks and unintended consequences of advanced AI systems Establishing best practices for AI development and deployment to minimize risks Exploring the concept of AI containment and control measures Studying the potential for AI to cause existential risks and developing mitigation strategies
Explainable AI (XAI) Read More »
Preventing Unintended Consequences: Researching methods to anticipate and mitigate unforeseen negative consequences of AI systems. Robustness and Reliability: Developing techniques to ensure AI systems are reliable and function as intended, even in unexpected situations. Value Alignment: Aligning AI systems with human values to ensure they operate ethically and in accordance with human goals. Security and
The Cognitive and Linguistic Capacities of AI: Can AI stems understand and speak natural languages? Can they have beliefs? Can they have emotions? Can they perform intentional actions? Do they have goals? What are the similarities and differences between human and artificial intelligence? What is artificial general intelligence (AGI)? What is Superintelligence? Moral Status of
Date: May 3, 2024 (Friday) Time: 14:00 – 16:00 Venue: CPD 2.42, HKU Centennial Campus Speaker: Kathryn Goldstein, HKU Registration: https://hku.au1.qualtrics.com/jfe/form/SV_cwDbZRfzjnQDbAq This workshop will discuss plagiarism and ethical use of ChatGPT, but we’ll also work on building up prompt engineering skills. Examples of specific skills students will learn, using prompt engineering: Use ChatGPT as a
Using ChatGPT to Improve Writing Skills Read More »
Date: April 26, 2024 (Friday) Speaker: Prof Seth Lazar, Australian National University Chair: Dr Frank Hong, The University of Hong Kong Abstract: The recent acceleration in public understanding of AI capabilities has been matched by growing concern—from presidents, industry leaders, scientists, and the wider public—about its potentially catastrophic, even existential risks. But at the same time,
What, if Anything, Should We Do, Now, About Catastrophic AI Risk? Read More »
We are excited to announce the launch of a new collaboration with The Conceptual Engineering for Emerging Technologies (CEET) initiative. This joint venture brings together the Conceptual Engineering Network, the ESDIT Consortium, our AI & Humanity Lab, and ConceptLab HK. The CEET initiative aims to promote the application of conceptual engineering to transformations induced by
New collaboration with the Conceptual Engineering for Emerging Technologies (CEET) Read More »
Speaker: Dr Rachel Robertson, Hong Kong Baptist University Abstract: This talk argues that AI ought to support the acquisition and exercise of integrity: the alignment within and between our thoughts, motivations, identities, and interactions with each other and the world. I will argue that an integrity-centred framework is very useful for assessing cognitive, normative, and
A HCI/UX Framework for Virtuous AI Read More »
Date: April 5, 2024 (Friday) Speaker: Dr Kate Vredenburgh, London School of Economics Chair: Dr Frank Hong, The University of Hong Kong Abstract: Proponents of algorithmic decision-making have argued that the use of algorithms can reduce discrimination, against the baseline of human decision-making. One reason is the greater explainability of the models, or the ability to
The Limits of Explainability for Reducing Algorithmic Discrimination Read More »
Speaker: Prof Suzuki Takayuki, The University of Tokyo Abstract: Classical AI, which has been dominant until 1980s in AI research, regarded thinking as computation, that is, formal manipulation of symbols. Though this approach worked well in simple tasks, it has repeatedly failed in dealing with difficult tasks. One fundamental problem is that it is extremely
Updating Philosophy of Artificial Intelligence in the Age of Deep Learning and LLM Read More »
Speaker: Prof Jiji Zhang, The Chinese University of Hong Kong Abstract: It is now common to call biased algorithmic predictions “unfair”, and many criteria of fairness have been proposed to serve as side constraints in the training of predictive models for optimal predictive accuracy or as parts of the objective function to be optimized along
On the Alleged Trade-Off Between Accuracy and Fairness of Algorithmic Predictions Read More »