Microsoft’s leading engineers discuss how AI and humans create a perfect symbiosis
There are some people who believe that future technology could be stealing jobs that are currently performed by humans; however, humans and artificial intelligence (AI) machines seem to work better together. This information was shared by a recent study done by Eric Horvitz, chief scientist for Microsoft, Ece Kamar, Microsoft Research’s principal researcher, and Bryan Wilder, a student at Harvard University and Microsoft Research intern. This study, published earlier this month, evaluated the performance of both humans and AI machines on two different PC vision projects.
Eric Horvitz led Microsoft’s research programs from 2017 to 2020. The two projects in which the subjects were studied were about breast cancer metastasis recognition and galaxy categorization. While working on these projects, another AI model evaluated in which tasks humans succeeded, and which type of tasks in which AI systems performed better. According to this research, combining the training of humans and AI machines can enhance the galaxy classification. This means that AI can solve problems that are more difficult for humans to resolve, while humans focus on resolving issues in which AI system predictions had lower accuracy.
Even though there are several ways in which AI systems and humans can work together, the development of this technology wants to focus on training AI systems to be the ones deciding when a human presence is required and for what task. According to this research paper, segregating AI tasks and allowing humans to add integral context is what makes this method much better than just one of the parts doing it all. The current developments are still working in training AI learning models to supplement human strengths.