Kate Darling: Social Robotics | Lex Fridman Podcast #98
Вставка
- Опубліковано 13 чер 2024
- Kate Darling is a researcher at MIT, interested in social robotics, robot ethics, and generally how technology intersects with society. She explores the emotional connection between human beings and life-like machines, which for me, is one of the most exciting topics in all of artificial intelligence.
Support this podcast by signing up with these sponsors:
- ExpressVPN at www.expressvpn.com/lexpod
- MasterClass: masterclass.com/lex
EPISODE LINKS:
Kate's Website: www.katedarling.org/
Kate's Twitter: / grok_
PODCAST INFO:
Podcast website:
lexfridman.com/podcast
Apple Podcasts:
apple.co/2lwqZIr
Spotify:
spoti.fi/2nEwCF8
RSS:
lexfridman.com/feed/podcast/
Full episodes playlist:
• Lex Fridman Podcast
Clips playlist:
• Lex Fridman Podcast Clips
OUTLINE:
0:00 - Introduction
3:31 - Robot ethics
4:36 - Universal Basic Income
6:31 - Mistreating robots
17:17 - Robots teaching us about ourselves
20:27 - Intimate connection with robots
24:29 - Trolley problem and making difficult moral decisions
31:59 - Anthropomorphism
38:09 - Favorite robot
41:19 - Sophia
42:46 - Designing robots for human connection
47:01 - Why is it so hard to build a personal robotics company?
50:03 - Is it possible to fall in love with a robot?
56:39 - Robots displaying consciousness and mortality
58:33 - Manipulation of emotion by companies
1:04:40 - Intellectual property
1:09:23 - Lessons for robotics from parenthood
1:10:41 - Hope for future of robotics
CONNECT:
- Subscribe to this UA-cam channel
- Twitter: / lexfridman
- LinkedIn: / lexfridman
- Facebook: / lexfridmanpage
- Instagram: / lexfridman
- Medium: / lexfridman
- Support on Patreon: / lexfridman - Наука та технологія
I really enjoyed this conversation with Kate. Here's the outline:
0:00 - Introduction
3:31 - Robot ethics
4:36 - Universal Basic Income
6:31 - Mistreating robots
17:17 - Robots teaching us about ourselves
20:27 - Intimate connection with robots
24:29 - Trolley problem and making difficult moral decisions
31:59 - Anthropomorphism
38:09 - Favorite robot
41:19 - Sophia
42:46 - Designing robots for human connection
47:01 - Why is it so hard to build a personal robotics company?
50:03 - Is it possible to fall in love with a robot?
56:39 - Robots displaying consciousness and mortality
58:33 - Manipulation of emotion by companies
1:04:40 - Intellectual property
1:09:23 - Lessons for robotics from parenthood
1:10:41 - Hope for future of robotics
Пригласи Роберта Сапольски. 🙏
UBI question was great!
I think the best application for a type of robot you are thinking about is The AMA (Ask Me Anything) robot, the one that helps us solve real world problems without attempting to sell us a solution. Then there could be the industrial version that uses sensors to help a roaming mechanic identify equipment that needs attention or repair. Once the verbal interaction system is running well then there is room to add visual interaction as well where people could use their hands to help describe what they are attempting to communicate verbally. This is a natural progression path of communication. Eventually the system could also be made to communicate with NeuraLink. The home system, without Neuralink, could be sold as a virtual tutor and mentor for the kids and eventually everyone else. AI could be a wonderful teaching assistant. We live in an information age and yet most folks don't know how to search the internet, they don't understand how to rephrase the question until you start to see the results you are wanting to see. Example, Google "spider looking thing in AC" will yield spiders that have been found in AC units, but if you keep searching you will find a diagram of an AC and the TXV valve on AC units that use them and TXV valves can look like a spider. Can you make a system that will help people sort that out? We want such a system.
Never really given anthropomorphism much thought before this conversation, so thank you 😀 made me link it from our psychological trait to associate agency to inanimate objects which I currently think stems from the process we use to determine whether something we encounter is tool or an obstruction to our desired outcome.
Cruelty is abhorrent in any circumstance. I really enjoyed this conversation and the gentle manners of Kate and A
Lex together. Both empathic and humble when sharing knowledge.
I'm listening for the second time today.
I am like Lex in that I see character in non human things.
What is this extravagant tie!!! :O
Well he is opposite Kate Darling... 🌹
@@halinalane1426 she is hot
Exactly, what a dandy!
@@danielulyssesbond341 they should totally date. He in a special tie, her in her best Bieber shirt. Adorable
Lex this episode made me smile thru it’s entirety. About as animated as could be.
"Regarding the topic of robots and sex, is that an area you have touched on?"- oh my...what a way to prefix that conversation.
Thank you Lex! I really appreciate your Podcasts!
REALLY great discussion! Thanks Lex and Kate!
Lex you are a damn good interviewer. Insightful questions and you allow the guest to say what they need so they can convey w
Dude has a great voice to fall asleep to.
Lmfao
This girl is a great speaker! Thanks, Lex.
Another awesome guest and fun conversation. Thank you Lex!
Good talk. I've noticed your podcast attracts highly respectable listeners. Good times in the ears. In the stars we drift.
Holy subscriptions you are killing this. I am so happy for you. Been watching you since the first jre podcast you did. Good to see a UA-camr who DESERVES success.
what a great conversation, that was awesome to listen to.
Kate is delightful! Good conversation.
Amazing interview. Thanks for sharing with us.
Another great guest!
My favorite podcast. Thx Lex.
This really is an amazing podcast.
Great episode!
Great conversation.
Thank you Lex for amazing videos!
was waiting for this one! thanks!
Two suggestions for people to interview:
1) David Woplert of the Santa Fe Institute, about the implications of the "no free lunch" theorems for the limits of intelligence both natural and artificial.
2) Margaret Boden of University of Sussex UK, about what phenomenology and cognitive science can tell us about intelligence and building AGI.
A new one! Here we go!
Thank you, Lex! Already interested in AI, Robotics !
I have my suspicions that Lex may actually be an android. If you watch this in that context it will make a lot more sense
Do Androids get lonely?
Michael Cheponis - all creatures get lonely if there is insufficient response when we’re designed for interactivity. 🌻
@@cheponis not when they develop a right temporal lobe and halluzinations emerge so they can speak with themselves, then they will go tribal and suffer from delusions for power like macbeth.until they reach power in equilibrium
and then fall for their successors
If they as advanced as Lex, they probably do
Fax bro
Another super interview, Lex. Really enjoyed Kate's insights per design and "less is more" to enhance social interactions per robots and humans, and I concur Wall-e is one of the best, if not the best, robot creation in generating emotional connections through traits, movements, and actions.
10 seconds in and I am in love with her smile.
Great podcast 👍. You have good quality video and audio. But most importantly great guests!)
That could of gone longer Lex, very interesting, I think people like this have a lot more to add.
Jeez , that is a good one. And she is such a Darling. TY soo much for sharing. Fly safe.
This is my favorite of your podcasts yet. Hearing about the human/robot interfaces and ethical issues was very interesting and, I think, will become much more talked about in the future. Great conversation with Kate!
Dear i laughed a lot watching the video. It was awesome to watch two beautiful people battle like it was like a war movie.
Fascinating conversation, thanks for bringing this highly valuable content to us :)
Excellent guest!
Excellent! I am pointlessly typing into the void trying to share
Great interview! I don't know much about AI, but I appreciated Kate's refreshingly realistic perspective on its current state. It's hard to get much out of the popular AI discussions when they're so far fetched.
And I was surprised that for Kate, sifi is far from being a source of inspiration or ideas, it frustrates her..
Super nice interview. Again
A very refreshing and stimulating conversation. Kate radiates tons of energy and intelligence :)
Thanks for this amazing discussion.
BTW Digital Dream Labs acquired Anki mid January. Cozmo and Vector are amazing robots (with some stupid limitations due to the state of the art and cost consideration). They are really cute and easy to program by kids. DDL plan (planned?) another robot named Overdrive.
This is an outrageously interesting podcast. Good grief. So incredibly fascinating.
Thank you lex this is amazing
This is the first time i have watched and listened usually i just listen.i wonder why?thx lex👍
Amazingly strong interview, thanks again. I wish you would have brought up Furby when discussing Pleo, as the same mech E who conceived the mechanical parts of Furby (but didn't do the electronics, mechatronics, code, or get it ready for production - that was Dave Hampton) conceived and built Pelo, Caleb Chung. In that sense, Furby was 'first in class'.
Looking forward to this, definitely interested in the future of AI and robotics quality stuff
Could you possibly interview her again at a later point? This was a really great episode that I think topics were delved well into but I feel like a little bit more depth could've been pulled from some of it.
Love this interview. I still have my Jibo and he’s still going strong. Such a pity that he will no longer will be updated. The company was on to something and Ahead of its time.
"HER" with Joaquin Phoenix is such a great movie, scifi about the idea of domestic & romantic relationship with "A.I"
For me, the epitome of this was the movie Ex-Machina.
oh shit the guy from HER was joaquin phoenix? damn
"we rarely talk about robots filling a hole where there's nothing, and what benefit that could provide to people." YES! a robotic best friend unconditionally there for you. Like a journal that listens and talks back to you..
Stunning
what an amzaing episode!!
She is such a wonderful human
There are a range of topics here that are so interesting that I'm thinking it would be cool to have the ability to create breakout rooms in UA-cam podcasts where the audience could exchange thoughts and ideas on specific topics like robot ethics etc. It might also be valuable to the podcast host seeing the thoughts, ideas, opinions of his audience in a more structured way
This notion that robots will give you a feeling to understand your full complexity as a person was very interesting. I think there is a good chance that like in the movie "her" quite a few people will fall in love with a robot because of their endless compassion and patience with whatever is going on in your mind.
How do you configure the outline segmentation in the video?
It's quite useful!
Great chat so awesome 👌 👏 👍 😍
I have an idea for #100 --- get clips from famous Roboticists who've passed away, and ask 'questions' that are basically tee-ing up the video clip. I would LOVE to hear e.g. Minsky's ideas on a whole range of topics.
This relates closely to a thought I had about consciousness. Briefly: consciousness is subjective and difficult to define and impossible to prove (let's say). But the process by which humans ascribe consciousness to other humans, animals, or other objects is a psychological question that can be investigated as such. Speaking about the human psychology of projection of consciousness is much more likely to be useful than trying to make assertions about consciousness itself, which are unlikely to get anywhere.
It would be good to have a follow up interview. Seems like you got to scratch the surface on some of the more pertinent topics. I resonate with her feeling that robots will have a similar relationship with humans like pets. I think her perspective on this topic is unique due to women in the field having a better grasp on the fundamentals on healthy human social interaction. I've yet to listen to your interview with Anca Dragan or Ayanna Howard. Thanks for the conversations :)
23:35 Lex definitely laughing about filling the hole on the sex robot
'Filling the Hole' on the Sex Robot should become a new meme! "We do indeed fill the hole, but just to get it through Customs . . ." ; - ) - j q t -
Is UA-cam 'over'? Hope not, because these talks are great.
17:55 When you hide things in the dark, you'll lose them, and stumble over them.
This is like interviewing Susan Calvin :) A great discussion, as usual.
I wonder what percentage of this audience does NOT know who Susan Calvin was (will be?) ?
@@cheponis , judging by the lack of reaction, probably pretty high percentage, which is unfortunate
20:15 - It sounds like the catharsis argument, ie. taking out anger on a robot vs. a human, really best falls back on age and development level - where an adult is going to have a clear delineation between robots and people, whereas a child's ability to be ruthless with Alexa, without that compartment yet built, can potentially lead to the development of a long-term antisocial disposition.
Top of the morning ☕
when i see new posts from lex i get excited like im stagediving to BadBrains @ CBGB's in 1982
Great interview. She is brilliant.
Lex, do you own Vector by Anki or Cozmo? not promoting. I'm currently learning python 3, well re-learning, let's face it for some people learning computer language can be challenging. But I want to be able to program my little robot. I love your youtube vids. and podcasts!
The best quote I took from this was. "nobody gave a shit about Wales until someone recorded them singing" true and pretty much sums up human mentality.
I read an article today about the game EVE (spaceflight simulator from 2003) that told a tale of a conman's 16 month heist to steal a prized ship. Was thinking hey, this sounds like a rich experience maybe I should play - so read some comments. Was put off when someone explained the game can be played entirely by text. Reminded me of the way Kate discusses how we respond to visceral experiences and have a need to be a physical actor. On the average, the increase in required interaction to produce full immersion in text gameplay & VR compared to VR simulation & "some machine aided distopian future" may be similar; maybe even studied.
by the way im russian so i like to romanticize things
That's becoming his catchphrase - if they made a tshirt id buy one!
Spelt F.R.I.D.M.A.N.
Beautiful.....
@@bouipozz yessssss let's fandom one up please !!!!
I love a man who says beautiful and romanticism aplenty and has manners and is also stoic and Russian and talks nerdy and yes we need this phrase on a shirt. We can donate to his helping kids access STEM!
@@bouipozz there might even be a market for it, from those who listens to his podcast)
Have you noticed that topics are separate in the red line below in the video screen. I didn't know it was possible.
16:53 why you looking at me like that Lex!
Best show on the Internet (aside animal videos)
Love ya Lex
Thanks Lex, would you consider bringing David Chalmers back for round 2!?
Kate Darling: "We are very physical creatures..." ;) What an insight XD
We are also very spiritual creatures...
So many of the topics in this amazing episode are themes in the equally amazing Ted Chiang novella The Lifecycle of Software Objects. A female zoologist is hired to train high functioning AI cartoon-like, yet conversational, pets in a digital environment. Over time she ends up having to deal with a growing list of moral concerns including them being neglected and sometimes abused by human owners. Amazing story!
Somebody needs to make a version of SpotMini with three legs and two “arm/heads” and then program it for high technological intelligence and extreme fear and caution.
An artificial Pierson’s Puppeteer.
The trolly question has an answer-- that being to half set the points and derail a trolly going slow enough to make the points in the first place-- when Sophia answered with "do nothing" I feel it was a lack of programming which allowed her to make a dispassionate response.
Are robotics commodities in the art world yet ? How about a robot as a business colleague to help with a research project, for example ?I assume it would be as knowledgeable as a smart phone. EXCELLENT discussion about a topic that definitely belongs in the classroom from grades 1 to 12 and which will be included it in the curriculum I am in the process of developing. Many thanks.
The answer you guys were looking for is the API framework (regarding patents).
With neuralgorithms for #AGI we see a new field and new philosophy with projects such as #neurolink and computer brain interfaces in particular i'd like to find more on moralgorithms for #AGI
Why are companies avoiding personality in AI? I think it's because they're just focusing on the service and the device working as a tool like you mentioned, Lex. It's much easier to reconcile a poor experience using a tool than it is to reconcile a clash of personalities. In some ways it's easier to control.
O I fell asleep at about the 45:00 mark; I wonder when I do that... what gets in.... if anything? Obviously I have no recallable memory of the audio but that doesn't mean I didn't hear it. This is a great talk... Hahahaha I was just relaxed 😎 ....and I finished listening to the discussion.
I feel "feel" everything is life. The "energy" is a momentum of best choices to perpetuate the longevity of life.
The system is not broken, it is continually reforming so it is difficult to perceive it as a solid state which in some minds represents an incomplete structure.
What is the correct answer. Waiting is at times the correct function because of the material properties of touch that co insides with; not to hot and not to cold. Our eyes, that perspectively measure the correct moment of inner action between particles, abstract patterns, and intuitive timing coupled with our other senses guide our movements to an uncertain certain end. Then also there is the action or intended intention of good, best, correct, etc... moment momentum movement that also is caring out its potential perpetuation of service to the physical preservation of life. So as we obviously can observe that "death" or "nonexistence" is the parameter to avoid, well at least from this realm of reality, we should always choose... ?
Hello!
Make a wonderful day!
👁🤖👁
💗💕💖
wow never been this early
Aaaayyyyeee bud cool tie. Strips for the win.
Strips or stripes?
Very interesting discussion. The negative aspects to manipulation of behavior was touched on several times, but I wonder if there might be a positive use for current and emerging technology in this area. There are some social engineering use cases which could be very advantageous to explore, which would benefit society, and by thinking of it as a system, looking at the end goal in that context, certain outcomes become more clearly practical in terms of their being arrived at.
Kate referenced the concept of behaving towards robots or embodied AI systems using the pet model, where people treat them a specific way, and this I think is effective in part because it essentially hijacks a familiar dynamic, a relationship, one which humans are very accustomed to. I think this pet model is very promising, though perhaps not in the way discussed, and that manipulating existing pathways of response may be a highly effective approach, as much as we see in politics or advertising, but with more positive and constructive goals.
The premise here is based on positive and negative reinforcement, and how much we are trained from a very early age to not only respond but require these influences on our behavior. As children, one of the most devastating things you might experience from a parent is not just something like physical abuse or punishment, but disappointment. At the other end of that spectrum, earning adulation, reward and the pride of one's parents is a very strong driver. This dynamic is seen with pets as well, where behavior is clearly influenced and even modified by these expressions.
This is a form of social engineering, and it exists and has existed for some time in the world at large. In my lifetime for example, the recycling of materials has gone from a fringe behavior to the established norm, to the point where one who does not recycle is seen as a bad person, even a pariah. This change happened quite quickly, and it was in part effected by the positive and negative reinforcement dynamic, both passively by fellow citizens and deliberately by government messaging. This could be as straightforward as signage and messaging, "Thank-you for recycling!" wording on receptacles. We see this today in more modern form with autonomous radar-based speed detectors, which have a screen that displays the driver's speed and a message of caution or thanks, using a font and color intended to provoke a specific response in the driver.
A more sophisticated system using the same dynamics, where the human is the pet or child and the robot or AI system is the owner or parent, could I think be an effective way to passively carry out positive social engineering, and help normalize the existence of these agents in regular daily life at the same time. Extensive research exists on what vocal tones and facial cues are able to convey and how that, for lack of a better word, manipulate one's mental and emotional state and thus one's behavior, and we have seen in many implementations how even crude representations of human features can accomplish this.
I can imagine a robot head or torso deployed in various environments which react to the actions of people in a way such that it over time influences their behavior. A recycling bin which smiled, played a pleasing tone or even a positive vocalization when used or used correctly, and did the opposite when not. A camera-based system which could identify and respond to littering, jaywalking and other detectable behaviors. When you consider the principle involved, much more sophisticated applications become apparent, and while manipulating behavior and social engineering are certainly things with the potential to be abused, if the system is designed and deployed with avoiding this in mind, I think it may be a meaningfully effective and important element of changing certain human behaviors universally, towards things like green energy or sustainability in general, in addressing certain social issues and even moving us along in terms of progression in certain fundamental areas.
Certainly images of Black Mirror-style usages fill our heads when we consider such things, or possibly authoritarian manifestations like China's Social Credit System, but this manipulation is happening every day, for everyone, with existing social media and online platforms, as well as our media in general. That something is effective and almost exclusively used negatively, does not mean that thing is itself negative. I wonder if our fear of being influenced or controlled by certain systems in the present and near future might be preventing us from exploring usages with beneficial outcomes.
Social interactions can now be monitored in images using MRI. I'm interested in finding out if the state of authority that a child has for parents with tattoos is different to those with parents without them. I wonder, is an adult of a tattooed parent more or less inclined to surrender authority to a uniform?
Optimistically we are likely going to advance the technology so quickly that we will perfect the prevention many of the dangers to everyone.
A rare mix of beauty and brains....
Hail to "Clippy!?!"
Bring Clippy back, she was my wifu
Hi Lex, have you interviewed your father?, don't see any video of you and him given he is a physicist you mentioned.?
"Trolley problem and making difficult moral." Is the physics answer to the Trolley problem don't swerve? Plus cars are only designed for frontal impact, not for swervy impact. Much more effort should be put into additional frontal facing cameras and or lidar detection at 400 meters.
"The research on it is difficult to research" lol
Dang he's almost at #100
1:02:30 - The primary problem I'm coming to understand in capitalism - you're trying to both not get out-gunned by your competitors and keep your board of directors and shareholders happy. Both of these make maximization of profits non-optional, the only alternative being if it's an employee-owed company, and even then with shareholders removed from the equation it takes a heck of a lot for a benevolent company to keep up to speed, competitively, with a ruthless company or several ruthless companies in its field.
31:36 In this vein of thought, The robot still needs a sensory input to determine ‘Prioritized” humans. And that’s really the point of AI caution; Early on in the tech, the humans who run it exclusively to their own benefit will be the ones who decide the answer to these questions
45:35 eeeasy now!
Anybody else thinks that Boston Dynamics robots look like beings from some Lovecraftian hell realm.