Ayanna Howard: Human-Robot Interaction & Ethics of Safety-Critical Systems | Lex Fridman Podcast #66

[@lexfridman]

I really enjoyed this conversation with Ayanna. Here's the outline: 0:00 - Introduction 2:09 - Favorite robot 5:05 - Autonomous vehicles 8:43 - Tesla Autopilot 20:03 - Ethical responsibility of safety-critical algorithms 28:11 - Bias in robotics 38:20 - AI in politics and law 40:35 - Solutions to bias in algorithms 47:44 - HAL 9000 49:57 - Memories from working at NASA 51:53 - SpotMini and Bionic Woman 54:27 - Future of robots in space 57:11 - Human-robot interaction 1:02:38 - Trust 1:09:26 - AI in education 1:15:06 - Andrew Yang, automation, and job loss 1:17:17 - Love, AI, and the movie Her 1:25:01 - Why do so many robotics companies fail? 1:32:22 - Fear of robots 1:34:17 - Existential threats of AI 1:35:57 - Matrix 1:37:37 - Hang out for a day with a robot

[@AlistairAVogan]

Left this episode feeling particularly optimistic. Conclusion: at the very least, we’ll make great pets.

[@thiruvalluvar3880]

Im a Georgia Tech student, glad to see the our great professors being represented here!

[@lorenzoatzeni4712]

such a pleasure to listen to her!

[@kwisclubta7175]

Rayght? Like, she sounds so, like, professional? And, like, mature? Rayght? Such a great representation of, like, GT professors? Rayght? Like, she doesn't sound like a preteen dingbat at all? Like, rayght?

[@thiruvalluvar3880]

@@kwisclubta7175 damn, dude I actually never thought her language was childish.. she's a well accomplished professor, you know.

[@kwisclubta7175]

@@thiruvalluvar3880 Really? Jesus, how many times did she say "like" and "right?" And why does every statement she makes sound like a question? I'm not saying she's not intelligent or accomplished. Not sure why you felt the need to point that out. But she does speak like a damn teenager. If you don't see it then your bias is causing that. I'm from Georgia too and I used to DJ on Tech's 91.1fm radio station but my pride doesn't make me deaf.

[@thiruvalluvar3880]

@@kwisclubta7175 yeah, but at this point I think you're just nitpicking for speaking style. I didn't really notice it that much, but you're probably correct. I'm not good at picking up on speech differences.

[@vaibhavbangwal]

I have been a Ayanna Howard fanboy for quite some time now. Thank you, Lex!

[@AlistairAVogan]

The quality has been consistently excellent. Never feel like we’re returning to an old point. Thanks for keeping it fresh.

[@charlesdavidrobertson9783]

Another terrific interview. Ayanna Howard was clear, knowledgeable and very interesting. Thank you.

[@metabalancetv]

Excellent episode; Ayanna will carry the project forward effectively and with clarity and curiosity.

[@KabzieMusic]

Another fantastic guest, Lex. Quickly becoming my go-to podcast.

[@solarwind907]

Lex, I’ve been meaning to write and thank you. Your interviews are always great. You always seem to have great chemistry with whoever you’re talking to. You set a high bar for public service for the rest of us with your interviews. Sincere thanks and best wishes to you.

[@Angel-nk9dd]

This was an amazing conversation, really enjoyed the talk on the ethics of programming of autonomous systems and developer accountability when it comes to things like death and accidents

[@eyykendrick]

loved this woman! good lord its a nice change to have such an open person

[@Stwinky]

Awesome! Love when we get researchers from all disciplines

[@scottypimpin5901]

Listened to this twice already

[@BiancaAguglia]

The “blood on your hands” issue is a rabbit hole there is no indisputable way out of. Do Karl Benz or Henry Ford, through the impact they had on the car industry, have blood on their hands because of all the people who die in car accidents? Do inventors of electrical devices have blood on their hands because of all the people who die from the misuse or the malfunctioning of those devices? Do WE have blood on our hands because we are using cars and electricity (and therefore keep their manufacturers in business?) As I said, a rabbit hole that can leave anyone a bit unsure, if not downright uncomfortable, about some of their life choices. The wise approach is to, as Ayanna summed it up so well: 1. be constantly aware of the responsibility that comes with your gift, 2. do your absolute best in your work, 3. seek unbiased and trustworthy feedback about the quality and ethics of your work, 4. be brave enough and humble enough to correct course if needed. You’re working on self-driving cars because you’re trying to solve a real and meaningful problem. You’re trying to make lives better, maybe even save them. You have good reason to believe you’re on the right track. In other words, you have good reason to believe you’re not deluded: data shows that self-driven cars can be safer, overall, than human drivers. You care. You work hard. You give your best. To me, you have nothing to feel guilty about. I’m confident you won’t be sleeping well for at least a while. You’re too driven to give sleep the proper place it needs in your life. 😊But don’t let guilt be what keeps you up at night. Instead, let your sleepless nights be driven by curiosity, passion, and wisdom. This was another wonderful interview. People like Ayanna and you keep me hopeful about the future, inspired to work alongside those who are building that future, and proud to be a human being. 😊

[@jonaspiva41]

Humans live on the the lives of other organisms full stop. Eating plant and animal, moving around building houses etc. I think that Lex is a sophisticated enough thinker to allow a perspective like that in, extract it's potential value as well as accept and release it. Press on good man. Cooks have blood on their hands, doctors have blood on their hands, all humans have blood on their hands. Religiously it is "original sin," the sin of being human. The best we may hope and strive for is balance, not purity.

[@monikafl2267]

I like those questions about ethics, it's really important to think about it

[@jaymorris7296]

I would have to agree with Ayanna on the hypersensitive reaction with Autopilot. I find I'm a much more aware driver with it. I trust it, but I'm verifying it's doing what it's supposed to. Another great job Lex, I really enjoy your Podcast and guest. Keep up the great work!

[@SuperEssiEs]

As always Lex, this is a great interview. Thank you.👊

[@markgrandau752]

Great talk. I'm a trust theory junky. Trust theory is a division of Game Theory. About 10 years ago, I read a book that showed trust has multiple definitions that form a Bayesian network. Trust is the fundamental currency of any relationship between a person and another person, place or thing. Excellent talk.

[@jonaspiva41]

fucking cool.

[@benitadiop2703]

I enjoy the new perspective! Thanks Lex

[@MrN0tim3]

This is amazing!! Much respect.

[@blind_neighbourhoodNerd]

Thanks so much for this amazing content!! This has become my favourite podcast listening, while messing around in Unity.

[@TravisGarnett]

Thank-you so much for #THIS conversation!! Simply amazing....#KUDOS 👏

[@steveoca]

The classic Lex question. When do you think... " we'll have robots in our homes and Robots in our hearts? " Yeah, Love it :-)

[@normdeplume4106]

re "Developers being responsible for the death of a human" - this could probably be extrapolated across many, many products throughout human history. A person stepping in front of an autonomous vehicle are akin to kids choking on a Lego. There are any number of things in civilization that are designed to be benign and are largely helpful or beneficial, but that will have outlier cases where someone dies because of that same thing.

[@grantrichard6974]

Thought-provoking discussion on not all intersections have the same rules and the trouble they cause autonomous vehicles. Like humans, does the AI have a super-cautious mode for unfamiliar and complex situations? Or do we imagine an intersection specific AI to help - as in asking a passenger to be a second pair of eyes?

[@billbutler8560]

Very interesting lady. Thank you.

[@learnwithfish]

She has very good points here. It is almost impossible to not get these negative biases...

[@scottypimpin5901]

Your amazing lex keep it up!

[@gradyberton]

Great interview, RIGHT?!

[@seanfitzgerald4207]

42 minutes in.....suggestion of a bias bounty program to surface ethical issues in AI i like it!

[@polishfish]

I’d never heard of an Ethics Bug in training data, but it makes perfect sense! 20:04 was an interesting ethical problem as well

[@Heeroyui752]

47:00 A sign of the times that times are good and prosperous because there's drama, profound.

[@corylowe5557]

Lex, you seem extra happy during the into for this episode. Maybe that's just my observation, or maybe there is a cool story behind your smile, but i'm excited to watch this episode.

[@linchenpal]

Great talk... for the insurance part : actuaries developed in the past UBI insurance on demand as well as telematics...

[@richardvilla2303]

So great, thanks for this interview~

[@deadpoolbr0]

such a vivid conversation, like it a lot!

[@nonchalantd]

great guest

[@ogungou9]

Love is a medieval literary invention ... It's an illusion. There is attachment, affection, empathy+compassion ... 1:31:32: A hammer and a person working together against a nail ...

[@rowenab.747]

Very interesting conversation.

[@cloudhidden00]

Thank you!

[@Hoondokhae]

best yet thanx

[@prenticedarlington2720]

The shortcut for autonomous vehicles is more dedicated (faster, non optical wavelength) sensors. Ideally, the road environment needs to be tailor made for autonomous vehicles, such as human excluded roads, special machine readable signs or clear lane delineation for the vehicle sensors (or virtual rails) etc.

[@EhKurd]

57:00 this is the type of alienation that makes engineers dread these ethics people. You know she's referring to all the nerdy tech people she couldn't get along with.

[@shehabmohammad6511]

Brilliant

[@winstonthomas4576]

Amazing. Your most human interview

[@thefonsotube]

2010 addressed Hal's ethical constraints. Recall the Doctors conversation with Hal at the end.

[@MrBlue-km8qv]

Robots are products you pay for. For perfection with robots i think of obedience and adaptability. if ball rolls into the street and a kid chases after it into the street the car had better have had predicted what could happen and it should have slowed down.

[@claytonbarnette]

We will have to turn up Autonomous Vehicles (AV) one lane at a time while we improve AV. Each municipality, county or state will have to govern these lanes much like they govern Truck, Toll and HOV lanes but at the same time manage them like they are railways or airways. In some places, the interstates are 4 lanes wide, once most areas reach 3/4 of the total lanes being autonomous; neighborhoods, cities, and people will adapt AV much the same way we've adapted to our phones. Deaths with AVs will happen just the same as deaths-by-selfies and normal accidents happen. Once we get the right leadership, mindshare and resources in place, daily life with Autonomous will become the norm.

[@simonsuh1733]

yay

[@DaveWard-xc7vd]

Is Dr. Ayanna Howard a member of Mensa?

[@achunaryan3418]

My personal favourite robot is the one who programmed lex

[@sunnos999]

The reason the carrent firms have higher prices for people under 25 is that that the human brain is not fully developed until 25-30 years of age and it is the part where adult characters as empathic ability, impulsecontrol and consequence thinking. The frontal lobes is the part of the brain we are talking about. Generally it takes longer for males to get it developed (and for some people it isn´t enough developed ever, and a "hidden disability", without medical help). So it hasn´t really to do with age but the biological age of them (the developement of the brain). But the question is if we at all should be in traffic if we are that much worse at that stage of life? It is one of the reasons that most of the crimes are done by those under 30 or by people that started in crimes before the age of 30 and got that bad habit/lifestyle they can´t break without help from society (if you repeat thing enough times it gets into a habit and the longer you do it the harder it is to break as known this far). Imagine how much "stupied" stuff you did before 25 that you now would never do and shake your head about.

[@erinvader9257]

💛💛💛💛💛💛💛💛💛💛💛💛

[@justchaz.]

Musing here. What was the cost benefit of missing out on talent like these for Slavery, Jim Crow, Segregation and institutional limitations of all sorts that this nation decided to engage in. How much better off would Blacks be if that equity had been allowed to stay with them rather than transferred to whites.

[@warrenbarker369]

And not just Blacks but society in general!

[@hankmmxviii2640]

Please have Robert Miles on next

[@blacklabelmansociety]

Hey Lex, I’m a huge fan of your work. Truly inspiring, it makes me reflect about life, technology, AI and much more. Hope someday you will receive Jordan Peterson or Bill Gates as guests.

[@thiruvalluvar3880]

Im a fan of Lez,but JBP isn't really an AI dude..

[@Baldebinos]

Several times You ask if HAL 9000 (in the movie 2001) was wright / wrong / good / bad / moral / immoral considering the decisions he took. Have You seen 2010 (the movie, of course)? The answer is there!

[@thefonsotube]

The trial of Lieutenant Com. Data: "Measure of a Man"

[@tomamerman7699]

Right?

[@joshryan4387]

I'm kind of shocked she thinks that having to read a manual implies bad design. How else are people supposed to understand the capabilities of the system, especially one that's potentially very dangerous and so different from anything they've ever used? "It's full self driving!! No need to read the manual."

[@jonaspiva41]

Maybe she knows it's dangerous to hide important safety shit in the manual? The good designers can allow for intuitive uses or users, in fact that's what makes them good designers.

[@markgrandau752]

People don't come with manuals. I agree it's a current short coming mostly due to "legal". Smart devices robots should walk you through their capabilities using the human interaction interface. But that's just my opinion. lol

[@eldyy9328]

Wait so what if a cop pulls somebody over in a Tesla who was texting while not driving? Is that still a ticket?

[@jonaspiva41]

probably, sense you are required to maintain focus on the road in order to intervene. Do you let your spotter text when you are lifting at the gym?

[@matespanjic8580]

With the growing robotics, machine learning, 40% of jobs as we know it disappearing by 2024 and as much as 90% by 2030 there are a lot of questions popping up, such as: 1) Where the future jobs will be created? What kind of jobs these will be? Are we going to be better of as a whole? Are we all going to do coding, selling 3d printed hamburgers made of human flesh or such? 2) Are we going to be Free at all, and what kind of freedom it will be? 3) How the consciousness idea will be handled, as well as the decision making process resulting from it? Whose ,or rather which idea of consciousness will be adopted? How that relates to the freedom, especially individual freedom? 4) I can SMELL a strong scary sense of SINGULARITY all around me. Feels like these new world in creation is just a predictable EXTENSION of the Mind CONTROL systems going rapidly DIGITAL all over the world. 5) Economics can not be ignored. In that regard one has to understand fundamental economic principles, as well as the history of the economics backed by the RELIABLE figures. One has to understand the 'commons' (past and present), interest rates and profits, QE or money printing, wealth transfer, etc etc. Talk about Negative interest rates , which are real by the way. Manufacturing capitalism vs Speculative capitalism? Horrors of socialism are real too, and costly. 6) So far, the most relevant questions is 'Who Owns and Rules AI'?Who decides where we go next, and more importantly how did we get to this point, and who should be deciding where are we going from here? 7) Possibly even more important Question is : Are Some People Trying to Play Gods? Perhaps, those should be first human, just to get some sense what it is like to be human? 8) Technology is good and necessary. We do need and invite the technology , but the type that won't bring SLAVERY and BONDAGE and SUBMISSION of any kind. This is way the GOOD PEOPLE of this world have to do everything to protect the Dignity and Sanctity of human life. 9) Back to SINGULARITY. It appears to invoke a DOGMA like state, that sometimes take generations to get rid of it. Unfortunately , it always leaves roots deep down. We do need the Explosion of fresh ideas and the renaissance of Imagination. We need the new ENERGIES to start creating the new World. The World where hope and happiness abounds. Something that DRIVES us away from this CULT LIKE STATE. There is a plenty energy for it, it is just a matter how we channel it in a best possible way. 10) We have to clarify RATIONALLY and FULLY the distinction between TRUTH, POST TRUTH, Fake Truth, VR, Staging before we move forward. Otherwise we can not move forward in any significant manner. We also need full freedom of speech and relevant platforms available, so we can get the maximum of the ideas. (including the silly ones). We need to learn from the failures of the previous and the current economic systems and start devising some type of fairer economic system. It can not be done overnight. The sooner we build that the more successful we would be in managing AI or whatever you call it. That would be some type of INCLUSIVE SYSTEM, where we all CONTRIBUTE and AIM for betterment of all. That would be that big PARADIGM SHIFT , new frontiers. Do not want Mr Musk in a few years to lobby for the compulsory Neuralink (doing well on the stock markets) or such. 11) Ayanna says we do not have classical definition of love. So she feels free to help design or redefine what the love is. Suppose all the concepts of being human are in the workshop as we speak. She also knows there would be people left behind beyond the edge of hopelessness, not even capable to play 'Hunger Games' since they would not have any strength left. Tough luck. Show must go on? That strata of society would not fall in love with the super intelligent self learning robot, especially with the ones patrolling the dense populated smart cities bein built right now. 12) Hey Lex and Ayanna there is another question: Can you Code a Compassion? Ayanna: yes, it would be accurate under the current presumptions. We could stuff up but we would keep adjusting, until the AI takes it over, under some kind of human supervision. Then Lex adds a smartie: How we would know it is a real compassion or not? Ayanna: AI would decide, since AI would ultimately know best.

[@j.hanleysmith8333]

A black woman makes it into robotics so you know she is a fucking genius.

[@DanielGarcia-gs9sv]

This reminds me of the old saying " we're playing God again "

[@kwisclubta7175]

Like, this is the best podcast ever? Right? Like, I really enjoy, like, the guests and subject matter? Right? But sometimes you, like, get someone, like, insufferably annoying? Rayght?

[@YouLookSoHappy]

Take a good look at yourself and question your motives. Why post multiple comments like this? Here is your other comment: "Rayght? Like, she sounds so, like, professional? And, like, mature? Rayght? Such a great representation of, like, GT professors? Rayght? Like, she doesn't sound like a preteen dingbat at all? Like, rayght?" (or both) reaction to her being on the show, and as a result you're unjustly picking her apart for the way she speaks.

[@richardvilla2303]

@@YouLookSoHappy Yeah, this is juvenile behavior and more annoying than someone using a filler word like "um" "uh". And an illogical take, it's not unprofessional to use these common words in this way, it is a dialectic trait of many parts of the US.

[@ubachiuchendu5385]

Get a job. She has a PHD in robotics, what are you doing?

[@burkebaby]

I have been a Ayanna Howard fanboy for quite some time now. Thank you, Lex!