BOSTON, MA -- Dr. Jennifer Golbeck, a computer scientist with a diverse educational background from the University of Chicago, Harvard and the University of Maryland presented the opening keynote address at Total Solutions Plus (TSP) on Monday, October 28, 2024 in Boston, MA. Specializing in Artificial Intelligence (AI), social media and malicious online behavior, Golbeck began by teasing that the discussion will be both enlightening and somewhat alarming -- akin to watching a horror movie but assured the audience that no real harm will come from the information shared. She set the stage for an hour-long talk with time for questions and answers at the end.

Golbeck outlines several points of AI: 

The Threat Narrative in AI

There is a prevalent narrative around AI being a potential threat to humanity, often propagated by those in the technology industry themselves. Golbeck suggests that this narrative serves dual purposes: to attract venture capital by emphasizing the technology's power and to distract from more immediate ethical and practical issues with AI. She debunks the fear of AI taking over by humorously highlighting AI's current limitations, like its inability to distinguish between certain objects or produce sensible color names for paint.

Predictive AI and Its Implications

Golbeck delved into the history of predictive AI, using an anecdote about Target predicting a teenager's pregnancy through her buying habits. This part of her talk outlines how AI processes data to make predictions, often without understanding the 'why' behind its conclusions, leading to ethical considerations. Golbeck explained how AI, by learning from historical data, often replicates societal biases, as seen in the COMPAS algorithm used in judicial systems, which has been criticized for racial bias.

Data Collection and Privacy Concerns

Golbeck also addressed the extensive data collection practices that feed AI systems, using examples like mobile apps listening to conversations or tracking location even when settings are turned off. She points out the privacy implications, especially with sensitive data being used for marketing, as illustrated by her dog receiving mental health awareness ads from CVS based on his prescriptions.

Generative AI and Its Limitations

Finally, Golbeck introduced generative AI like ChatGPT, explaining it is good at mimicking human speech but not at factual accuracy or understanding. She shares cautionary tales of misuse, like lawyers using AI for legal briefs with fictional citations and discussed potential legal and ethical pitfalls in deploying AI without human oversight. She concluded by advocating for a cautious approach to AI integration, focusing on augmenting human capabilities rather than replacing them, and stressed the need for transparency and bias auditing in AI applications.