Robots for the Rest of Us Podcast

What Does a Humanoid Robot Do? I Asked a Humanoid Robot

I am at ICRA – the IEEE International Conference on Robotics and Automation, aka 7,000 roboticists from all over the world converging to discuss every robot-related topic on Earth. More posts are coming, but for now, day-of, here is a quick podcast interview (the first with video) with, and about, one of the robots I saw in action today – Ameca, by Engineered Arts [https://engineeredarts.com/]. After many an online video, this was my first time interacting with an Ameca robot in the (gray rubber) flesh. It was a different experience than watching a video, in much the same way a conversation in-person is different from watching a video of other people talking. I don’t mean to suggest that the robot is the equivalent of a person or spurs the same thoughts and emotions. But it spurs some emotions and reactions, outside of conscious control, either innate or conditioned (which is not a subject for this post), that usually come up in a human conversation. Yet I walked away cautiously optimistic about my ability to cope with people-like objects, on screens and in 3D life, as they proliferate. There is little danger of anyone confusing an Ameca with a real person – not just because of its inhuman skin and eye color, but because (in this incarnation) it did not act like a person, with its eerily calm voice, unnaturally even intonation, and somewhat pat responses. (Others’ experience will vary, by design: Ameca has a number of available personas, and Engineered Arts sees it as a platform for developers to put their own AI into. I interacted today with one persona, at the Engineered Arts booth in the exhibitor’s hall.) Still, the clarity of the boundary between human and robot (which Ameca brings up) gives me hope that people will not be easily fooled or manipulated into treating robots as people. In fact, I find myself wondering if human-ish robots may actually be safer on that score than AIs, simply because AIs that “sound human” come to us over the same media as real people do. The medium that is a text from a sophisticated AI doesn’t look or feel different than does a text from me (though I am still better at the content part of the message). But a robot body can’t offer that sense of “this is the same as with a person.” The illusion – that machine and human are the same type of being – is hard to maintain. So my guess is that conventions will soon develop for dealing with robots that act human. These conventions will be based on human-to-human norms, but they won’t be the same. Maybe, as some have suggested [https://pmc.ncbi.nlm.nih.gov/articles/PMC9795935/], those norms will include built-in distancing gestures by the robot, like Bertolt Brecht’s “distancing effect” in the theater [https://www.linkedin.com/pulse/ai-insights-focus-epic-black-boxes-art-breaking-spell-wemyss-k6iif/], so the audience doesn’t lulled into complacency and daydreams. For example, at Kodaiji temple in Kyoto, the Mindar “talking statue” robot [https://www.kodaiji.com/mindar/] was deliberately designed to show off a mechanical body under its human-like face. It’s a reminder that the robot is not human. Ameca looks rather similar to Mindar, actually. Engineered Arts plans to give Ameca functional arms and hands (good enough to pick up a chess piece) and, later, legs it can walk on. But the company has built in other distancing effects – including that non-human-looking skin. [https://www.uniladtech.com/news/ai/reason-advanced-humanoid-robot-gray-skin-222768-20250124] I think people will continue to feel sure that robots are not people – that the machines are, instead, representations of people. Like puppets, video game characters and characters in novels. (This is an idea that a number of thinkers have arrived at lately, from different starting points and in different disciplines. I’ll have more to say about this soon.) So go my day-of thoughts about the experience of Ameca. As with most encounters with a complex robot, it was much less cartoonish, and more complex, than what I imagined. See what you think. Get full access to Robots for the Rest of Us at robots4therestofus.substack.com/subscribe [https://robots4therestofus.substack.com/subscribe?utm_medium=podcast&utm_campaign=CTA_4]

The Work AI Can't Help With

Why hasn’t workplace artificial intelligence made life easier for everyone? Some surveys see a vast majority of employees saying AI [https://www.forbes.com/sites/bryanrobinson/2024/07/23/employees-report-ai-increased-workload/]adds [https://www.forbes.com/sites/bryanrobinson/2024/07/23/employees-report-ai-increased-workload/] to their workload, for a variety of reasons (having to check what the AI does, having bosses who expect increased output). And the extra stress of AI adoption is often worse for women, says leadership consultant Julie Donley [https://drjuliedonley.com/]. Why? Because AI is often oriented toward things men tend to focus on (like maximizing efficiency at all costs) and not the jobs that fall much more on women (for example, the emotional labor of office politics, and the running of both work and home lives). Because women are rightfully more worried about being judged by office culture — for using AI (“did I cheat?”) but also for not using it (“do I look like I’m not keeping up?”). Because the sped-up expectations that AI can create put even more pressure on women at work, even as they still do more than their share at home. I spoke with Donley about these and other ways AI adoption can make for more burnout and exhaustion among women workers. But this isn’t a gloomy conversation. She believes AI can be made to work for us all, if it’s adopted in a way that fosters human flourishing. We talked about her new book, Leading at the Speed of People [https://www.amazon.com/Leading-Speed-People-Strategies-Fast-Paced/dp/0976560577], in which she discusses how leaders can shape AI to make work more humane, not less. Give us a listen! As always, please note the transcript is AI generated and could contain mistakes. Get full access to Robots for the Rest of Us at robots4therestofus.substack.com/subscribe [https://robots4therestofus.substack.com/subscribe?utm_medium=podcast&utm_campaign=CTA_4]

How to Manage AI Therapists, Robot Friends, AI Ghosts

Most people agree there’s something wrong with an AI pretending to be a human being. And especially wrong about an AI telling you it’s a human therapist. In fact, a California legislator introduced a bill [https://www.vox.com/future-perfect/398905/ai-therapy-chatbots-california-bill] the other day to ban AIs from doing that. It stemmed in part from an incident where a mental health platform had users thinking they were getting counsel from humans when in fact the messages were written by GPT-3. [https://arstechnica.com/information-technology/2023/01/contoversy-erupts-over-non-consensual-ai-mental-health-experiment/] Here’s the thing, though: The patsies (I mean, users) actually rated the responses quite highly – until they learned they were machine-made. This “empathy paradox,” as Anat Perry calls it, revealed an aspect of life where human beings don’t just want a product (like good advice); they also want human effort. Perry, a neuroscientist who is professor of psychology at the Hebrew University in Jerusalem, has a lot of insight into why this is so, and what it could mean for AI as a tool in therapy. We talked about that in this podcast. It’s a rich topic, because many instances of machine pretense aren’t as clear-cut as the fake-therapy incident. Fact is, any sophisticated AI or robot gives people double vision: You know it’s just a machine but other parts of your mind feel that it has emotions and thoughts. So how do you manage that experience? How much can you control it, and how much help do you need from society, in the form of new laws and norms? Now that we have AI therapist apps [https://lotustherapist.com/], AI romances [https://boyfriend.myanima.ai/], AI friends [https://heyamica.com/] and even AI ghosts [https://projectdecember.net/], people should be talking about these matters. So check out our conversation! Or have a look at the AI-generated transcript. Bonus: Some relevant links to things that came up: “Considering AI-driven therapy: When does human empathy matter?” [https://s3.ca-central-1.amazonaws.com/assets.jmir.org/assets/preprints/preprint-56529-accepted.pdf] (a recent paper by Perry and her colleagues) “Generative Ghosts: Anticipating Benefits and Risks of AI Afterlives.” [https://arxiv.org/abs/2402.01662v4] Thoughtful paper about the artificially resurrected. “Eternal You” [https://www.justwatch.com/us/movie/eternal-you] Great new documentary about ordinary people reckoning with AI recreations of deceased loved ones. “An overview of the application of artificial intelligence in psychotherapy: A systematic review.” [https://www.termedia.pl/An-overview-of-the-application-of-artificial-intelligence-in-psychotherapy-A-systematic-review,46,54543,1,1.html] (Links to a pdf) Get full access to Robots for the Rest of Us at robots4therestofus.substack.com/subscribe [https://robots4therestofus.substack.com/subscribe?utm_medium=podcast&utm_campaign=CTA_4]

Before AI Permeates Our Lives, We Need a New Deal for Privacy

Data scientist Wes Chaar was at home a few years ago, working with the tools of his trade, when he had an “aha!” moment. He realized the information he was looking at would let him predict not only what people would buy and how they’d vote, but also how they’d feel about their choices. Those people might not have shared those feelings with anyone, but Chaar could foresee they’d be there. How are people supposed to defend their privacy in the face of that power? All those user agreements we click without reading [https://www.theguardian.com/technology/2017/mar/03/terms-of-service-online-contracts-fine-print], and those assurances that our data is “anonymized” before it’s crunched, are no protection at all. It’s a more urgent and pervasive problem now, as AI and robots enter take more tasks in people’s lives (whether they want them or not). Knowing that Google or Amazon can figure out so much about you from your searches on their sites, imagine how much more they’ll know as they’re connected to AIs that write your emails, give you vacation ideas and track your sleep and mood; or to robots that clean the house and entertain the kids. Chaar believes we’re at a crucial juncture. Now, before AI and robots are woven into everything, is our collective chance to reshape the terms of the bargain with all those data hungry services that companies are so eager for us to use. In his new book, Data Independence [https://www.amazon.com/Data-Independence-Reclaiming-Privacy-Evolving/dp/1642259756/ref=sr_1_1?crid=L6G6PZNTDI1W&dib=eyJ2IjoiMSJ9.lCIG4_KoT9_oyiOvKK_WNiDh455zp_vPAXPD8QvZ4ZWU_AKh_b7GL3TYdOm3r3bFP5sKjXTthS7Ir1cP8Ft4djW971wVbcq7h0OvDFIdico.YpBxRKE50pmfV1vtNX0J5Z0c_qM_mSK4Pknwgb-u5dY&dib_tag=se&keywords=data+independence+reclaiming+privacy&qid=1736884550&s=books&sprefix=data+independence+reclaiming+privacy%2Cstripbooks%2C73&sr=1-1], Chaar proposes a new deal for information: A system for keeping control of information even as we share it so that our AI and robot assistants can do their work. We talked about this proposal, as well as the overarching issues — what is privacy? How can we get around a paralyzed political system to enact changes? What’s the strat for getting around the opposition of companies that like today’s wide-open data-plucking regime? And a lot of other vital topics. Give a listen! As always, the transcript is AI generated and may contain some errors. Get full access to Robots for the Rest of Us at robots4therestofus.substack.com/subscribe [https://robots4therestofus.substack.com/subscribe?utm_medium=podcast&utm_campaign=CTA_4]

Sometimes You Just Need a Robot to Scratch That Spot You Can't Reach

We’re an aging society, with a shortage of young-ish people to do the work, paid and unpaid, of helping others get through life. (Immigration crackdowns will make this shortage more acute.) Inevitably, robots are coming to help us in the home. Maybe they’ll be 5-foot, 8-inch tall imitation people, as Elon Musk predicts [https://www.thewrap.com/elon-musk-tesla-optimus-robot-explained/]. But arms, legs and torsos are expensive to build, complex to operate and kind of frightening/creepy. So in the years just ahead home robots will likely be simpler devices. They’ll be designed to do useful jobs in their own robot way, not imitate a person doing the job in a human way. But how do you make robot that’s actually helpful, that people will want to use, that doesn’t cost a fortune? What’s the right balance between versatility and simplicity? Who gets to decide what “helpful” means? How do engineers – always eager to do cool never-before-seen things – find common ground with us civilians, who want something that will work without confusing us, frustrating us or scaring us? Charlie Kemp [https://charliekemp.com/] has been thinking about those questions for a long time. Formerly a professor at Georgia Tech, he’s now CTO at Hello Robot [https://hello-robot.com/], which makes Stretch, 50-pound mobile manipulator that has been used for many more purposes than its creators imagined when they started. These include farming tomatoes, running physical therapy exercises and finding missing stuff. The background visual for this episode is another use: It’s a photo of Henry Evans, who is quadriplegic, using a Stretch to play with his granddaughter. Users have also taught engineers that they sometimes want a robot to do much simpler things than the robot makers thought of. For example, being able to scratch an itch via robot can give a paralyzed person more autonomy, and makes for one less request for help to a human caregiver. We talked about this and other lessons from real-world uses – and about the near and farther future of assistive robots. Give a listen! As always, the transcript here is AI-generated and may contain errors. Get full access to Robots for the Rest of Us at robots4therestofus.substack.com/subscribe [https://robots4therestofus.substack.com/subscribe?utm_medium=podcast&utm_campaign=CTA_4]