Poonam Chaudhary
IEEE Member
What does it mean for AI to be a tool or an assistant, rather than a replacement?
“Some people excel in programming, others thrive in content writing, data analysis or creative endeavors. With AI, this diversity can be leveraged to bridge gaps, allowing individuals to enhance their abilities without the burden of being all-rounders.”
Chenyang Lu
IEEE Fellow
What skills help people work with AI more effectively?
“To work effectively with generative AI, people should learn to interpret AI outputs to understand their implications and limitations; leverage AI insights into decision-making processes; and provide feedback to enhance AI models.”
Eleanor Watson
IEEE Senior Member
How can we train people to understand the limits of AI?
“The articulate qualities of chatbots lead us to overestimate their capabilities. Their inherent randomness can spew genius or nonsense based on an algorithmic dice roll. It's important to recognize that there is therefore often as much work in checking the work of AI than in creating it oneself.”
Euclides Chuma
IEEE Senior Member
How can we make sure AI models stay current given the changing nature of the world?
“Time changes everything, including personal and collective opinion. We need to ensure that these changes are reflected in AI models. We human beings have to create methods to ensure that the results of AIs are aligned with our ethical, cultural and social values at that moment.”
Karen Panetta
IEEE Fellow
Why is it important for humans to be able to explain AI decisions?
“AI can be used to screen job applicants or evaluate job performance. Simply stating that the AI determined something based on factors that can’t be explained to a human is not appropriate. We need to understand what kind of biases are in the data because the AI will reflect these biases in its decisions.”
Santhosh Sivasubramani
IEEE Senior Member
What is it about being human that AI is not likely to ever replace?
“Human empathy and intuition. It will be impossible to recreate these. In healthcare, AI can’t replace the human touch in examinations, even if it uses state-of-the-art sensors. In education, a human teacher encourages critical thinking, something that can't be replaced now or in the future.”
Cristiane Agra Pimentel
IEEE Senior Member
How can AI systems adapt to new and unforeseen situations outside of their training data?
“AI systems can adapt to new and unforeseen situations using strategies such as reinforcement learning and adaptive AI. These strategies enable systems to learn and improve from experience, adapt to changes in data and the environment and handle situations that are statistically similar to their training data.”
Kayne McGladrey
IEEE Senior Member
How do hallucinations – false information generated by large language models – limit AI’s usefulness?
“We need tests of the ability of AI to provide accurate information. That’s not a one-and-done thing. AI learns from data which is periodically refreshed. If we can’t test whether an AI is accurate right now, it really becomes difficult to trust it in specific applications.”
