In early December, NLL Senior Researcher Dr. Chris Dede gave a keynote speech at Empowering Learners in AI 2022, the third year of this international conference. ELAI 22 focused on the theory, models, and practical impact of artificial intelligence (AI). The keynotes and panels addressed questions such as:
- What are the theoretical underpinnings of AI adoption in learning settings?
- Which existing theories of learning provide insight into learning when AI is an active partner?
- Which models of adoption of AI are changing the quality of the learning experiences of individual students?
- What is the practical impact of AI on education?
- What is the infrastructure needed for AI adoption in schools and universities?
- What are the ethical considerations of theory and model adoption?
- How do schools, universities, and corporate learning departments assess and evaluate the impact of AI? What is the economic measurement of impact? How are the learning gains measured?
Chris Dede’s keynote is entitled “Intelligence Augmentation via Artificial Intelligence: IA rather than AI.” Its description is:
Science fiction often portrays “intelligence” as involving complementary roles of reckoning and judgment. For example, in the Star Trek series, the judgment and decision making of ‘Captain Picard’ is enhanced by the reckoning skills (such as calculations, administration, and computation) of the android ‘Data’, a machine without human capacities like emotions. This partnership is Intelligent Augmentation (IA); the human and machine work synergistically together to be greater than their individual abilities.
While many forecasts chart an evolution of artificial intelligence (AI) in taking human jobs, more likely is a future where AI changes the division of labor in most work roles, driving a need for workforce development to shift towards uniquely human skills. This forecast implies that learning knowledge, skills, and dispositions for work should increasingly prioritize capability building of human judgment, applied wisdom, and decision making—at the expense of developing some reckoning skills that AI will assume. Moreover, machine learning (ML) could be used to help “engineer” learning, by applying evidence-based strategies to the continual re-design of performance-based simulation experiences to optimize their effectiveness and efficiency. This will enable developing diagnostic/formative longitudinal assessments of judgement that complement our current high-stakes tests centered on reckoning.
The keynote is based on a NLL brief on Intelligence Augmentation: Upskilling Humans to Complement AI co-authored by NLL staff Chris Dede, Ashley Etemadi, and Tessa Forshaw. Dede’s 1 hour keynote is available for viewing.
The keynote goes beyond the brief in forecasting three major impacts of generative AI on society over the next decade. Generative AI apps such as ChatGPT and DALL-E 2 call into question which tasks previously done by people will now be done by machines, such as descriptive writing and creating detailed works of formulaic art. Dede’s talk predicts that, with wise use of generative AI, society can benefit in at least three ways:
- AI-based “what if” modeling may improve some types of human decisions
In the 1970s, spreadsheets revolutionized business planning by enabling agile financial modeling. Now, generative AI can predict “what if” outcomes of natural phenomena even when no scientific study has been conducted to produce an estimate. For example, given current data AI could forecast the likely seawater level in various parts of the city of Boston in 2050, based on ocean effects due to mid-level climate change. Such use of generative AI can enable detailed planning without costly and time-consuming initial studies. However, AI cannot accurately forecast when people would take action to mitigate these effects, because human behavior is much less predictable than the consequences of natural phenomena.
- Human creation of innovative artifacts may be enhanced by AI providing “baseline” material
People can produce richer, deeper narratives if they build on descriptive writing generated by AI; for example, repurposing description into a compelling story drawing on motivational factors important for young adults. Similarly, people can produce more complex and creative art if its formulaic aspects (e.g., a border of identical, detailed flowers) is generated by AI. In sculpture, robotic generation of a stereotypical three-dimensional form can be a foundation for human creative expression through additional carving. Additional benefits for human creation are likely to develop as AI progresses.
- Improved negotiation, collaboration, and empathy in the workplace
AI can now aid in teaching complex skills such as negotiation. Just as airline pilots can master handling challenging situations using flight simulators, so human beings can now practice sophisticated skills in working with others through immersive simulations, as illustrated by companies like Mursion. Research has shown that, done well, the social-perspective-taking involved in these experiences can increase empathy, advancing diversity, equity, and inclusion.
A learning experience from Chris Dede and Ashley Etemadi about human occupational skills in the era of machine learning and generative AI will be part of the NLL practitioner workshop series this March.