{"612329":{"#nid":"612329","#data":{"type":"news","title":"How Will People Teach Robots in the Future? Georgia Tech Researchers Identify What People Prefer in a Robot Learner","body":[{"value":"\u003Cp\u003EImagine ten years from now anyone being able to go to the store and pick out a robot housekeeper. They bring it home and it begins to organize and clean the house, except it\u0026rsquo;s not exactly how the homeowner wants it. They need to teach the new robot how they prefer to have the clothes folded, dishes put away, and, yes, the trash taken out.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EGeorgia Tech Machine Learning Ph.D. student \u003Cstrong\u003ESamantha Krening\u003C\/strong\u003E and Associate Professor of Aerospace Engineering\u003Cstrong\u003E Karen Feigh\u003C\/strong\u003E published work on human preferences for how to teach AIs in their new paper, \u003Ca href=\u0022http:\/\/journals.sagepub.com\/doi\/abs\/10.1177\/1541931218621371?journalCode=proe\u0022\u003E\u0026ldquo;Characteristics that Influence Perceived Intelligence in AI Design.\u0026rdquo;\u003C\/a\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026ldquo;This whole area is really fascinating to me because it explores how people want to teach, and I feel like we very rarely ask that question,\u0026rdquo; says Feigh.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EAI or autonomous agents typically learn one of two ways: the critique method, where they are explicitly told something is done correctly or not, or the demonstration method, with a human physically teaching an action. Depending on the skill being taught, agents are more receptive to one method over the other.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EKrening and Feigh developed the Newtonian Action Advice method as another alternative. This new method incorporates verbal human advice for actions with reinforcement learning in order to improve human-agent interaction.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026ldquo;Historically, people have always asked questions like \u0026lsquo;How can we teach better?\u0026rsquo; or \u0026lsquo;How does the learner want to learn?\u0026rsquo; But in this case, we need to be asking, \u0026lsquo;How does the teacher want to teach?\u0026rdquo; when scientists are designing the robot learner,\u0026rdquo; says Feigh.\u003C\/p\u003E\r\n\r\n\u003Cp\u003ETo answer this question, the duo surveyed people on how they felt about training robots using the critique method versus the action-advice method and came up with several key attributes that people desired in a robot learner.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ECompliance with input \u0026ndash; \u003C\/strong\u003EDid the agent do what it was told?\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EResponsiveness \u0026ndash; \u003C\/strong\u003EHow quickly the agent learned the skill\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EComplexity \u0026ndash; \u003C\/strong\u003EWas it simple for the person to teach the agent the skill?\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ETransparency \u0026ndash; \u003C\/strong\u003EWere the humans able to understand why the agent made its choices?\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ERobustness and Flexibility \u0026ndash; \u003C\/strong\u003EThe agent\u0026rsquo;s ability to correct mistakes and learn alternate policies\u003C\/p\u003E\r\n\r\n\u003Cp\u003EThe results showed that people tended to be drawn to the action-advice method because they could see the agent making adjustments and learning in real-time, versus critique, which does not provide the same immediate, visible results.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EAs researchers create technologies, such as household bots, that are managed or trained by people with little or no computer science experience, keeping these attributes in mind is important Feigh says.\u003C\/p\u003E\r\n\r\n\u003Cp\u003ESmart devices have already introduced AI agents into the home, so the idea of a domestic robot doing the dishes exactly how the homeowner wants might not be too far off.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cem\u003EFeigh and Krenig will present their findings at the \u003Ca href=\u0022http:\/\/cms.hfes.org\/Events\/Annual-Meeting.aspx\u0022\u003E2018 Human Factors and Ergonomics Society (HFES) International Annual Meeting\u003C\/a\u003E in October. \u003C\/em\u003E\u003C\/p\u003E\r\n","summary":null,"format":"limited_html"}],"field_subtitle":"","field_summary":"","field_summary_sentence":[{"value":"Georgia Tech researchers study people\u0027s feelings and perspective on teaching robots tasks into account as they build robots for consumer use."}],"uid":"34773","created_gmt":"2018-10-04 15:18:55","changed_gmt":"2018-10-04 15:18:55","author":"ablinder6","boilerplate_text":"","field_publication":"","field_article_url":"","dateline":{"date":"2018-10-04T00:00:00-04:00","iso_date":"2018-10-04T00:00:00-04:00","tz":"America\/New_York"},"extras":[],"hg_media":{"612328":{"id":"612328","type":"image","title":"Robots could be built based off how the owner likes to teach. ","body":null,"created":"1538666024","gmt_created":"2018-10-04 15:13:44","changed":"1538666481","gmt_changed":"2018-10-04 15:21:21","alt":"","file":{"fid":"233090","name":"Picture1.png","image_path":"\/sites\/default\/files\/images\/Picture1_2.png","image_full_path":"http:\/\/tlwarc.hg.gatech.edu\/\/sites\/default\/files\/images\/Picture1_2.png","mime":"image\/png","size":351295,"path_740":"http:\/\/tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/images\/Picture1_2.png?itok=EBQ40j48"}}},"media_ids":["612328"],"groups":[{"id":"576481","name":"ML@GT"}],"categories":[{"id":"8862","name":"Student Research"},{"id":"135","name":"Research"},{"id":"153","name":"Computer Science\/Information Technology and Security"},{"id":"152","name":"Robotics"}],"keywords":[],"core_research_areas":[{"id":"39501","name":"People and Technology"}],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[{"value":"\u003Cp\u003EAllie McFadden\u003C\/p\u003E\r\n\r\n\u003Cp\u003ECommunications Officer\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eallie.mcfadden@cc.gatech.edu\u003C\/p\u003E\r\n","format":"limited_html"}],"email":[],"slides":[],"orientation":[],"userdata":""}}}