Radiolab

Navigate
Return Home

Breaking News

Thursday, July 27, 2017 - 10:09 PM

Simon Adler takes us down a technological rabbit hole of strangely contorted faces and words made out of thin air. And a wonderland full of computer scientists, journalists, and digital detectives forces us to rethink even the things we see with our very own eyes. 

Oh, and by the way, we decided to put the dark secrets we learned into action, and unleash this on the internet. 

 

Reported by Simon Adler. Produced by Simon Adler and Annie McEwen.

Special thanks to everyone on the University of Southern California team who helped out with the facial manipulation: Kyle Olszewski, Koki Nagano, Ronald Yu, Yi Zhou, Jaewoo Seo, Shunsuke Saito, and Hao Li. Check out more of their work pinscreen.com

Special thanks also to Matthew Aylett, Supasorn Suwajanakorn, Rachel Axler, Angus Kneale, David Carroll, Amy Pearl and Nick Bilton. You can check out Nick’s latest book, American Kingpin, here.

Support Radiolab by becoming a member today at Radiolab.org/donate.

Guests:

Matthew Aylett, Nick Bilton, Hany Farid, Durin Gleaves, Ira Kemelmacher-Schlizerman, Jon Klein and Steve Seitz

Tags:

More in:

Comments [74]

Tom Lum from NY

If we want to fake audio of Trump or Obama all you need is someone who can pull off a decent impression... This episode felt like one enormous scapegoat blaming technology from the real issue of fake news... the fabrication of fake news isn't the issue at hand, it's the belief of it, the lack of people seeking out multiple sources and confirmation and critically thinking. There's no holy validity or platonic truth to audio clips or video clips, and there never has been, just as with print.

Aug. 21 2017 10:00 AM
John from Carolina

Dilip, you were correct about their truth right up until they pulled the episode because someone complained about the truth. Now they've edited the truth to fit their audience and become just like all the other fake news organizations. It's shame they didn't stand up for the truth.

And Flip, it concerns me that both sides will be misrepresented.

Aug. 20 2017 09:28 PM
Jennifer

Do you think apps like FaceRig have the ability to capture facial expressions and voices for later use like these digital impressions? Could be that everyone using it is pretty much voluntarily offering up data for use.

Aug. 19 2017 07:24 AM
Angelica Dawn from San Diego

What will happen to voice actors?! Will that job cease to exist with this program? Will you instead have to sell the rights to your voice?

Aug. 18 2017 01:05 AM
Dilip Kondepudi from Winston Salem, North Carolina

Radiolab
I was riveted to my iPhone as I listened to your “Breaking News” episode. Thank you for making your listeners aware of technology that will have great impact on our world. I have the following thoughts I would like to share..

- We are moving into a world in which Truth will become a very valuable commodity. Just as material technology has resulted in the production of processed foods or as the author of “In Defense of Food”, Michael Pollan, calls it “food-like edible substances” or fake food, technology of the digital world has begun to produce recorded-reality-like videos or fake recordings.
In the material world, the consequence of fake food is the growth of groceries like Whole Foods and a growing market for more expensive unprocessed and organic foods, because real food is what we need. It matters. People still enjoy fake food, but most know (and some don’t care) that fake food not healthy. Similarly, in the digital world, radio and TV that can maintain their credibility for reporting facts and unprocessed reality (we understand some editing is essential) will grow in value because truth matters. So programs like yours will become even more valuable and important than they are now, and listeners like me will be willing to make larger contributions than they are now — I certainly am, and I have made my contribution today. I am not worried that the honest journalists will cease to exist, but I know that it will cost more to support them.

Jad, Robert and all who work at Radiolab, there is money to be made. What you produce will grow in value. Don’t lose this opportunity to become rich. Put on your entrepreneurial hats and figure out how!!!!

Aug. 16 2017 10:47 PM
Jeremy Leyland from Earth... but moving

In theory.... could you use this technology, in conjunction with AI to re-create a loved one that died? Like combined with machine learning, video like in star wars, and voco... you could bring someone back... technology, this could be like harry potter portraits kind of thing.

Aug. 16 2017 04:30 PM
Tiffany J. from Orlando, Fl

https://www.instagram.com/p/BXxx5vSjh1a/

I heard this podcast about fake news then saw this a few days later. Lol, cute but the whole idea that this can be placed in the wrong hands....not so comical!

Aug. 16 2017 07:56 AM
Leah from Irvine

This was some seriously belligerent investigating. After the Adobe commercial, you feature a few men and their technologies respectfully, but then aggressively grill the woman technologist and demand this non-American answer for American fake news issues writ large? You actually recorded her umm-ing like that after putting her on the spot, and actually aired it? No other technologist was given this disrespectful attitude or even remotely challenging questions throughout the show, it was painful to listen to.

Of course technologists should be mindful about what their designs engender, but radios were also used to fool the masses -- this is not a new idea or phenomenon. Further, this new technology has plenty of other implications outside American fears of "fake news" that cut across cultures / intersections that y'all don't even bother touching on. This ep was a fear-mongering, lazy, self indulgent, offensive, and v disappointing.

Aug. 15 2017 12:38 AM
Huxley Ford from Panama City FL

Couldn't help but notice... You skipped Barak Obama and went straight for George Bush. Nice... I still love the old episodes, send me an email if you ever come back from the crazy left...

Aug. 13 2017 06:35 PM
Benjamin Reaves from Silicon Valley, CA

I'm quite glad to see this technology available - just like photoshop, it teaches us to be wary of believing what we hear too quickly. This technology seriously works - not only Adobe but other companies and universities not mentioned on Radiolab for example lyrebird.ai (no I won't work for them)

Aug. 10 2017 10:59 PM
Dr Schaefer from Phoenix

I'm not concerned about political operatives making fake videos of Obama or Trump. I'm more concerned about the FBI using something like this so they no longer have to deal directly with Muslims, rather they pick random brown dudes on facebook and make videos of them "confessing" to hatching terrorism plots that the FBI makes up, then the FBI claims credit for stopping these "terrorist events" that were never going to happen in the first place http://www.kansascity.com/news/local/crime/article135871988.html

Aug. 10 2017 04:44 PM
TjS from NYC

Sorry this is off topic but I am curious what the song at 47:13 min is?
Does anybody know?

Thanks!!

Aug. 08 2017 11:45 PM
A R J from WashDC

It could be the future of print media once the electronic media is completely fouled.

Aug. 07 2017 03:42 PM
Fubeman from Washington State

Well, if Jad and Robert (and I love them) actually finished watching the VoCo demo, they would have learned that Adobe is also working on implementing a watermark system to make all audio that was manipulated by VoVo easily detectable before they plan on releasing this. So there's that . . .

P.S. Also, the video that they used for their "Fake" demo was quite horrible and sooooo easily discernible as quite fake. The mouth movements looked like a bd puppet show, the blurring and pixelization around the mouth was horrendous and the timing was off by a mile. Sorry. If the video aspect of this is going to scare me, it is going to have to up its game - quite a bit.

Aug. 06 2017 07:52 PM
Linda from Seattle

I appreciate Radiolab's generous, inquisitive approach to their subject matter -- whatever that subject matter may be. Science fills me with wonder, so it's not too difficult to be awed by a science-focused episode. Others such as the hunting and conservation ep really elevated the discussion and explored the nuances and realities in ways I hadn't considered.

This episode felt a bit foreign to me. I wasn't filled with awe and wonder. I wasn't presented with a contextually rich exploration of a technology. And while I'm certainly accustomed to gotcha, soundbite journalism, Radiolab usually eschews such tired tactics. This wasn't the case with the Ira Kemelmacher-Schlizerman clip that perturbed other commenters. Perhaps she's never considered the ethical implications, but perhaps she simply didn't have a polished, Supreme-Court ready response. Not everyone is articulate when put on the spot or as gifted a communicator as, say, Robert Sapolsky.

I haven't listened to the next episode. The blurb positions it as somewhat redemptive. I look forward to hearing it.

Aug. 06 2017 07:18 PM
DeGroot

It was mention in other comments, that the ability to convincingly alter still photos is a bit of a template for how this may roll out. I've thought for some time now that photo manipulation is one of the factors pushing people to be suspicious of everything, not just photos. People know that photos can be manipulated so were much less trusting of a photo that shows us something we're not quite prepared to believe.

We've now also had years of people pushing the idea that the media is manipulating us and to not believe that other media source.

This developing video technology will only add to people's tendency toward confirmation bias. What is left that can reach across all these entrenched opinions and say, "I am truth?" More and more the lay person is going to go with their gut on what sounds plausible and reach to the media source that confirms this.

We are in deep doodoo as a society.

Aug. 06 2017 11:54 AM
Jad Abumrad from my anxiety.

Freak Out Now!! An evil scientist named Tim Berners-Lee has written a program that let's people publish ANY words that can then be seen anywhere in the world! With paper books you know something is true but with this new system TRUTH WILL DIE!

Aug. 06 2017 11:53 AM
Claire from Washington, DC

I don't this this is cause for too much panic. Humans have long told and written lies...created illusions. We've always believed what we hear and read to the degree we trust our source. We would benefit most from developing more powerful, objective ways to determine the trusworthiness of the source. The trick is developing an algorithm that isn't (and isn't perceived as) politicized. We're already suffering for the lack of such a thing.

Aug. 06 2017 11:34 AM
L from MI

I am incredibly disappointed in you guys for this one. The futureoffakenews clip is laughably bad - the inconsistent audio, the cartoonish facial movements, the stock camera flash sound effects, everything. I cannot honestly believe anyone found it convincing. Based on their reaction, I'm seriously questioning the judgment (or, sad to say, the integrity) of Jad and Robert here.

Aug. 05 2017 11:53 PM
Sylwia

This technology brings a new level to the ability to people to catfish others online by deceiving someone they are speaking with someone else via video, giving a greater false sense of security.

Aug. 05 2017 11:32 AM
Paul Clarke from Evanston, IL

Fake news is, and will always be about trust, not technology.

Methods will be developed to tell if a video or audio clip has been falsified, and even if these are expensive, you will always be able to check with credible news organizations for their take on a story you see online. The real problem is getting people to trust the people with the truth, and not the fakers.

Aug. 04 2017 06:42 PM
Chuck B. Ans from Spokane

I like a story on fake news where the interviewer pretends to be surprised by the things he's learning from the person he is interviewing.

Aug. 04 2017 02:43 PM
Gabriel from Mexico

The disregard for the societal impact on the part of the scientists that are developing these technologies is staggering. They’re just playing with they toy questions and techniques to make them work, but they don’t care at all about what this means for society as a whole. Not to mention their opaque connections to Google and Facebook. I can see why you left that long silence after Dr. Kemelmacher-Shlizerman’s struggling with such a simple question as “Are you afraid of the power of this?” and her response along the lines of “I’m a computer scientist…people should work on that” Why not the scientists in the first place? they are developing these techniques and they should be the first to be aware of what they mean to society.

Aug. 03 2017 12:00 PM
Ben from Parker CO

When you are motivated by fear everything can look like a monster. The most disturbing thing I heard in that story was the new layer of polish getting applied to the jackboots.

Aug. 01 2017 09:33 PM
sleepingtiger

The site referenced in show is
Futureoffakenews.com

Aug. 01 2017 09:32 PM
David from Los Angeles, California

Wow. You guys caught a lot of flak for approaching the story the way you did. I think you did a great job of opening the eyes of non-tech people to this.

We had this very same discussion 30 years ago when Photoshop just came on to the scene. We were able to manipulate picture in Photoshop on a Mac and it blew our mind and wrenched out gut.

For the next 10 years, when you tried to fake a photo, the "customizations" were obvious. Partially because we couldn't scan (no digital photography like today) and work at the resolution necessary to hide the work. There were blurs around the seams, there were lighting and color issues, the final rendering of the image was pretty bad. And, we'd throw our hands up and say "one day".

That day arrived about 5-7 years ago when you could easily work at a 10 Megapixel level - you could take the picture at the resolution, work at that resolution, and then "shrink" the image back down to a reasonable resolution for moving it.

And now, it takes a forensic expert (like your guest) to tell the difference. Perhaps video manipulation will have the same time frame??

Regardless, within those 30 years, the term "Photoshop" became a verb. Not because so many people use the tool - but because everyone knows the ability is there for something you see to not be true.

And once we humans know of a think like that - we will always be ready for it. After all, that's what we humans strive to do - we look for patterns and oddities, explore them, understand them, so that we know to take them into account the next time we encounter them.

When the signal says "Walk", but the light is red. What's the first thing we do? Look around to corroborate. What are other lights telling me? What are the other people and vehicles doing??

When you get a text from your love one that doesn't "feel" right, you corroborate. It's coming from them, and it uses the same words they use. But its just "off" somehow. You call them.

So, yes, it is scary to see that someone can fool me. But now that I know it's there, I won't be fooled the next time I see a video that doesn't sound, look, and ACT like the president.

And maybe that's why you aired the story to begin with.

Aug. 01 2017 05:13 PM
Britten from Nevada

Idiots...why are the smartest folks who make things that everyone know is going to be used for the wrong purpose incompetent. To have the woman scientists just say "duh, don't know" when pressed about how this software is going to be used has to be the dumbest person I know with a IQ of 150+. I hope she take to bed that software designed by her and others have no real good purpose and ultimately will be used just to harm another human...good job.

Aug. 01 2017 04:52 PM
Message in a bottle from Brasilia

I believe that tools like these will continue to evolve, and what really has to change is our minds and the biased ways people tend to see the world. Education will have to make media literacy a priority. We have to start to educate people to think outside the box, or outside the red and blue boxes, and all the other boxes. This is the most challenging mission for educators of the future. It looks that there is a lot of science trying to make advance in what machines can do, but very little is being done to change our minds. There's a lot of irrationality in humanity, we really need to evolve.

Aug. 01 2017 01:04 PM
Danielle Drabeck from Minneapolis, MN

As a scientist I am constantly concerned by the impulse of folks to pose the solution of scientific censorship to the problem of evolving technologies that require ethical and legal changes as well. This approach has never been successful, and never will be successful because science proceeds across the world when technologies become available, whether we censor it or not. Holding back scientific process is like trying to contain a cup of water in a washcloth. The only thing that ever results form this is other groups doing it first (case in point, the nuclear program, the space race, stem cell research, IVF... etc, etc.. ad nauseam).
Let us strive to push for strong and thoughtful programs in philosophy and ethics, and let us also invest in ways to encourage collaboration between thoughtful philosophers of science and ethics and scientists. For the love of god, please let us start recognizing that science outreach to ADULTS and POLITICIANS is important, and will facilitate sound ethical choices. These are real solutions. These are the types of solutions that prepare us for the technologies of the future. The temporary band-aid of censorship in the face of innovation has only ever left societies to face an inevitable new future blinded and unprepared.

Aug. 01 2017 11:59 AM
Devin from Iowa

I was into this story and thinking yeah, that is a concern that people can make these fake videos and pass it off as real, but also thinking that we've kind of already been through this kind of thing--this just makes it a little easier.

The biggest surprise was after seeing the fake video with fake audio. That was awful. Why were the people of Radiolab spooked? Seriously? That was so obviously fake and I find it hard to believe that anyone might think it was real. Real let down. This story kind of seemed like there wasn't any real story here and you guys tried to turn into a story. I really enjoy listening to Radiolab, but this story was pretty disappointing.

Aug. 01 2017 10:18 AM
Chelsea Boyle from Berkeley, CA

At first I was quite anxious after hearing this, but then I thought about all the misinformation and lack of basic critical thought already rampant now. It happens amongst all walks of life, particularly on social media.

I'm a liberal person in an educated liberal bubble, but misguided information gets repeated to me _all the time_. All it takes is one confident-sounding person saying something surprising yet plausible and it gets repeated and rehashed all over the internet. Then the sheer numbers of repeated statements mimic credibility. Very few people ever bother to do the most basic google search on something they've read, especially on social media, even when they don't recognize the source as being credible.

I guess what I'm saying is, using fancy technology to create misinformation is actually almost overkill. Due to a lack of habituated basic critical thinking and healthy scepticism amongst the whole population, all that's actually required are some convincing-sounding words. I assume, though, that curated sources of information will be able to adequately take into account the possibility of falsifyed video and audio, should these technologies become common.

I guess _my_ biggest worry is actually cyber bullying - people using the technology to create defaming videos that are embarrassing, despite known falseness.

Aug. 01 2017 02:51 AM
David from South Carolina

Whatever benefits this provides the world the liabilities will surely outweigh it. But it's out there, and now we'll have to deal with it. Yet another fruitless arms race to occupy human time!

Jul. 31 2017 11:20 PM
Michael from NYC

Of course the prospect of easy, convincing fake videos being misused is alarming.

I think though, that there are a few things to keep in mind. First, the futureoffakenews.com clip is actually pretty crude. Obama's head keeps changing shape like a partially filled water balloon. I also very much doubt that the Voco demo was completely legit -- if it were being used in field conditions on data and edits that had not carefully screened to produce an ideal result, it would likely have sounded more like the robot presidents. It may be a very powerful tool, but I doubt very much it will be as easy to use as the demo made out. These tools are not quite there yet. Doubtless, they will continue to improve.

On the other hand, is puppeted video really a game-changer? As the Radiolabs folks pointed out the gunman whoshowed up at Comet Ping Pong pizza didn't require video, or really any evidence whatsoever, to set him on his course. On a more benign note, an acquaintance of mine recently shared (on social media) a photograph of Ewan McGregor in costume as Obi Wan Kenobi asserting that it was a picture of Jesus. People eager to accept things uncritically will continue to do so, video is just another item to accept or not.

When the time comes that people really can generate convincing fake audio/video with easily accessible tools, we will all need to evaluate the sources, and use our judgement to decide whether to believe it, just like we already have to do with still images and words. We just need to get over our delusion that video is somehow more innately trustworthy than other kinds of media. It never was to begin with --videos are often already altered in less high tech ways, or even simply taken out of context to convey a completely false message.

My feeling is that this is something to look out for, but less worrisome than the widespread state of ignorance that leads so many to accept fantasy as reality.

Jul. 31 2017 10:06 PM
Skep Tic from LA Westsiiiiide

How will I ever believe future Radiolabs after this. Alas, Audio/Video consumption has just become acts of faith, not facts.

Jul. 31 2017 08:11 PM
Paul from Paso Robles

I have no doubt that this technology is coming. There is a lot of money and power at stake here and it's way to tempting to have the ability to sway people to part with their money or to confirm or shape their political views.

No matter what fake news would be generated, the responses could be:

Side A: Our experts have analyzed the video and determined that it is genuine.
Side B: Our experts have analyzed the video and determined that it is fake.

Who believes what?

Even now, it's so easy to fool the masses.

Jul. 31 2017 04:18 PM
Sam from Bonn, Germany

I liked and listened to this podcast because it is an intriguing topic, but this topic is not fresh. There was a conspiracy podcast on Youtube about this months ago. And that was a conspiracy podcast.

Radiolab, I thought, is not a place for conspiracy podcast. Where is the science? More science! Gimme more SCIENCE, not just abuse of technology stories/scare topics.

If you want to go into abuse of technology, consider also the abuse of ignoring technology. Excellent example: the German auto-industry (and other auto companies) and how that effects all of Europe and the world.

I like science. Maybe someone can recommend a more scientific-geared podcast for me?

Jul. 31 2017 10:53 AM
Richard from NC

Listened to this episode this weekend.

Along the lines of VoCo look at what Google is doing with DeepMind: https://deepmind.com/blog/wavenet-generative-model-raw-audio/

Jul. 31 2017 08:11 AM

I'm very relieved. Both the audio segment of modified speech played during the podcast as well as the altered speech by Obama are ridiculously amateur looking. This is highlighted best in the video when Obama says "golf". The sound is clipped and obviously not part of the original. I know several video editors who "sentence mix" much better without the aide of speech recognition software.

The technology will certainly develop over time, but for now there is no real threat in mistaking fake audio/video for what it is. I would certainly like to play with Voco when it is available.

Jul. 31 2017 06:57 AM
John Morgan from Portland

More information does not make it easier to fake news, it makes it harder. Information has been faked countless times for countless purposes through history. It is currently much harder to fake news and information, and it will probably remain harder to fake news and information, than it was when sketches substituted for pictures, written accounts substituted for video, etc.

This counterbalancing fact which is indeed common knowledge to any college educated person found no place in this story.

The way this story may be most alarming, and therefore the way this story may be most engaging, valued, shared, re-listened to, etc. is to ignore contrary information and arguments, and maximize fear of consequences in the listener.

Dedicating some portion of this show to cogent, educated, respected and prepared people with a contrary opinion to all this alarmism, is all it would have taken from the many intelligent and decent people of RadioLab to overcome media bias and help re-establish trust in mass media. But they didn't.

That's what's scary. That so innate is the human drive to grasp the brass ring, our own minds quell pangs of hypocrisy, ambivalence, and doubt. We cannot see, even while recording a show on the future of fake news, the systemic tendencies of competitive media to distort information in pursuit of the biggest story possible.

Jul. 31 2017 01:42 AM
Roi James from Austin, TX

Does anyone really believe that politically biased news organizations would not use this technology to manipulate their audiences? Organizaitons such as FOXnews and Breitbart who apparently intentionally misrepresent facts and proliferate "alternative facts" in support of their political and ideological agendas. Not to mention the truly rogue, and purely vindictive media sources on the internet who seem to want to bring systems down just because they can. This software is an authoritarian state's wet dream. I heard of this tech before the Radiolab segment and felt ovewhelmed at what is coming down the road towards democracy and free societies. I don't know how we are going to be able to combat this when at present, intentionally misleading and untrue fake news stories which are written, not even backed up by video, have alarmed and activated segments of the population so as to divide the nation in two.

What upsets me the most is what appears to be Dr. Kemelmacher-Schlizerman's side-stepping the question of seeing the nefarious uses of the technology. Her evasiveness practically convicts her of the crimes to come. And they WILL come. If she had the courage to really look at that future, she might consider ways to protect the things that are precious in this world from this technology. She says she's building this technology so she can have a conversation with an Avatar mother. Well consider that one day, someone could use that technology to convict her and her mother of a crime they didn't commit by having them say something they never said. The most obvious focus of this tech is on political abuse by authority figures but I see police abuse through manufactured confessions that never took place. How can you say you never committed a crime when we (the police) have a video of you saying you committed the crime? The world is wide and waiting for the myriad forms of abuse this technology can and will take. But oh! I can't wait to talk to my avatar mom! Someone is going to get rich on this tech, regardless of the nations it brings down.

Jul. 30 2017 10:42 PM
chris LM from San Jose

This technology provides a lot additional opportunities for abuse. There is a corresponding technological solution, the incorporation of digital signatures, especially distributed ones like blockchain into recording devices. Trusted news would include the original media and allow for verification of the content. At least in terms of proof against digital manipulation and originating device.

Jul. 30 2017 05:54 PM
Mike from Madrid

I have to question Radiolab's credibility on the concern expressed in the Breaking News podcast. While I certainly agree that audio and video manipulation are serious and creepy threats to society, I consider Radiolab’s editing to be part of the problem. Can you honestly say that editors do not insert dramatic pauses into recorded interviews? Or have Jad re-phrase questions during editing (after the interview) to share your version of the story more directly?

Radiolab clearly has a significant amount post-production editing and sound effects. But on serious stories like Breaking News, this manipulation is not appropriate and can mislead listeners. I, for one, already lost some trust in Radiolab due to the very themes highlighted in Breaking News.

Jul. 30 2017 04:09 PM
Jacqueline

The scariest part of this podcast was Prof. Kemelmacher-Shlizerman's inability to answer basic questions about possible ethical concerns and washing her hands of any moral obligations here. It is not as if she should stop her work but at the minimum reflect on the possible applications and suggest policy or code of ethics that can introduce safeguards.

Jul. 30 2017 11:48 AM
Chris Gurin from Philadelphia, PA

This story arouses way too many inchoate thoughts to compose a sensible narrative,so I'll default to a grocery list of anxiety:

1.The myopic developers of this technology have unleashed a sort of cognitive terrorism we've all been dreading: an arms race of distorted reality and "alternative facts".

2. The greatest disasters in human history always arise from unintended consequences.

3. "When you see something that is technically sweet, you go ahead and do it and you argue about what to do about it only after you have had your technical success. That is the way it was with the atomic bomb."-
J. Robert Oppenheimer

4. The only way to stop something this awful is to unplug EVERYTHING: who wants to go first?

5. I think the doomsday clock has been tracking the wrong extinction event.

Jul. 30 2017 09:04 AM
Keith Duddy from Brisbane, Australia

Clutch Cargo was doing this in the 50s:

https://www.youtube.com/watch?v=UTQI_XjoSSY

...|<

Jul. 30 2017 06:48 AM
Steve from Pittsburgh

This may well be the scariest thing that I've ever heard. At this moment in history, the absolute last thing we need is technology capable of creating artificial spoken word and video of whomever we want saying and doing whatever we want. I don't think our democracy can stand it. I'm especially amazed at how the computer scientist that produced the video technology, Ira, appears to have never considered its possible malicious uses and seems totally unconcerned by them. What are they thinking? This is even more concerning to me than the potential nefarious uses of CRISPR - at least there are some regulatory bodies in science and its use requires some scientific training not possessed by the typical person. There are internet trolls all over the world that are going to have a field day with this. They've opened Pandora's box, and based on this interview, didn't even know it until now? I sincerely hope that they have a plan to regulate this technology because this seems like a "blood on your hands" type of moment.

Radiolab team - can you please do a follow-up with the people making this technology to press them on this point?

Jul. 29 2017 09:11 PM
T.A. Barnhart from Portland, OR

I'm disappointed you missed the most obvious protection, which is also the fastest growing way to get content on the web: livecasting. If there is an event with multiple livecasts – one on FB Live, one on Twitter, etc – and they match one another, we can know what actually happened. If something turns up that doesn't match the livecasts, then we know what's fake.

And people are doing live more & more. It's getting easier & easier. As long as something is livecasted, you'll have that record as your fact check.

And frankly, spreading false content has been done forever. It doesn't take tech; it just takes someone willing to lie & someone to believe that lie.

Jul. 29 2017 07:08 PM
You Tube from Uganda, Africa

This is why we need to stop worrying about global warming or politics, but focus on the explosion of AI and technology. I sincerely predict that by the year 2050, we will be able to affordable be able to upload ourselves into an external battery/simulation.

Jul. 29 2017 06:56 PM
Roman from Russia

Guys, the word/sentence mixing stuff is old news. People in the YTP community have been doing this for years! BY HAND!

Jul. 29 2017 06:39 PM
Galatae

Trying to look for positives, and I can see how some of this technology could be beneficial in healthcare, specifically for people who have catastrophic injuries that would affect their speech. Imagine if Hawking had 40 minutes of speech to turn into his own voice.

That being said. This is how wars are started. Say someone wants North Korea to go after the US and NOW they actually have the wherewithal to do it, what's going to stop them from thinking they have proof of some transgression they can just manufacture? It is no longer a matter of "If" it will happen, it's "When". These are really bizarre times we live in.

Jul. 29 2017 03:07 PM
zach from Raleigh, NC, USA

For a while now "digitally-signed" content has become more and more prevalent.

I am sure the authors of this piece got an ear-full on how digital content can be "watermarked". I do expect that unsigned content on "The-Net" will fast become a thing of the past. I am OK with the authors leaving this out since if they included it then the story would not have the same impact as it does, and I do think this technology should receive wide scrutiny.

Once digital content routinely has a 'watermark' then all social media sites will most likely only carry such content.

And, guess what? We could also extend this concept so you could trace-back everyone who shared/forwarded such content. And you could see just where it 'really' came from.

This will not eliminate "Fake-News" since some channel like The Faux News Network (Faux/Fos sp?) still could make up their own crap but the could NOT put 'words/expressions in/on someone's mouth/face'.

But when every single wad of digital content on the net has its own watermark and audit trail. Then it will be very straightforward to tell fact from fancy.

Jul. 29 2017 12:36 PM
David from Idaho

Ick... This was the first radiolab that I ever shut off half way through. They kept acting like this tech was one day going to used to cause great harm to society. Um... Nope! We're not all idiots thanks. People will know that this type of thing is possible to fake now and it'll only be used for jokes. Went way overboard with the communicating with ghosts thing. The uncanny valley still exists here and it's obvious.

Jul. 29 2017 03:05 AM
Daniel from Los Altos, CA

While the concerns of this piece are understandable, I think it undervalues how much context matters when people evaluate trustworthiness. It's not like this hasn't happened already - photo editing tools create edited images that are virtually indistinguishable from originals. Sure, this has led to some abuse by fake news, as well as increased skepticism that the images we see are real. But for the most part, society evaluates things well - they know when to cast a suspicious eye or call something out as probably Photoshopped, but it hasn't led to wholesale distrust of news photos. People know that images are easily manipulated, but still trust them, recognizing that where they come from matters.

Being able to do the same with audio/video takes this to a different level, but I'm pretty sure the consequence will be the same - people will be more skeptical with videos they see, but will still trust some, depending on the source.

Jul. 29 2017 03:03 AM
John

I'm pretty familiar with this technology, and I can tell you it will only get much worse. Imagine being able to synthesize an image or video of anybody you want saying anything you want, doing anything you want. Easily possible within 5 years. Only hope is some sort of cryptography to verify authenticity. Or maybe smash all the computers? Butlerian jihad!

Jul. 29 2017 02:43 AM
Spence from WV

A different spin on "fake news" from someone in local news: I could see an application once the quality is 100% where local news will use this so your local anchor person can be on multiple streams at once - or weather can be constantly presented by a likeness of your meteorologist so they "never need days off" and can literally be two (or more) places at once.

The cost savings to Broadcasters is immense. A likeness of Wolf Blitzer never needs to leave The Situation Room.

Or better yet - once they fix facial issues - a company can just make "the ultimate news presenter" saving tons of money. Build who you want and they never make anybody upset like Ron Burgundy!

Bad news for people like me: it's going to be easy to be replaced by this technology. Very easy.

It also brings up legal issues - most stations make "talent" sign away rights to their likenesses .... I would want a cut of profit if my "virtual self" was making money for the firm ... and the non compete ramifications are dizzying. What if someone like Lester Holt was to be hired by another network? He has to sit out a period of time - but if his former network had flawless technology they could virtually keep his likeness on air and performing for a long time until his non compete was up.

Think of "commodities" on air like traffic and stock reports. Those are already automated and a person just reads them. Computers kick out full blown stories for newspaper/Web for items like that on the fly- ... now add that text to produce a voice and matching facial image ... all done without human writers - bad for jobs for local media folks I say and a big can of worms.

May be 20 years away but sounds more like 10.

Amazing...

Jul. 29 2017 01:52 AM
Chris B

I really think Jad and Robert should re-examine this issue from a different perspective. Instead of running around like a couple of Chicken Littles, maybe they could look back at how the introduction of new forms of media or media editing technology has been treated in the past. I suspect that they'll find a lot of the same sort sky-is-falling rhetoric about things like photography, film, radio, audio tape, and digital images. These technologies have all changed society in complicated ways. Manipulation of recorded images or sounds have been around as long as we've had such recordings. Nothing they present in the episode (except perhaps that Jordan Peele clip, and that could have been done by any good human imitator) offers solid evidence that this kind of technology will make it harder to tell truth from fiction.

Jul. 28 2017 09:46 PM
Nicole from Miami, FL

That video in no way looked real. Either I'm greatly over-estimating the intelligence of the general public or I'm a hard person to "get one over" on, but that video was horrible! Looked like something a high school computer class could've put together. In no way was I duped into thinking it was really Obama saying that.

Jul. 28 2017 05:22 PM

OMG! I can't believe all those scientists that invented the internet 40 years ago. I'm sure someone was like "Once everyone can broadcast their own thoughts and opinions to millions how will people know the truth!?". "We must ban this technology, only licensed newspapers and TV stations should be able to reach that many people. If the average person has this tech it's the end of the world. Oh, I can't believe those scientists who invented the internet had no ethics. If they had they'd never have made it.

:rolleyes:

Fascinated show but there are MUCH bigger things to worry about than the latest Hollywood fx tech.

Jul. 28 2017 05:02 PM
Daniel T from Hillsborough NC

What I thought was the obvious question that wasn't addressed is why did the VoCo demonstration work so much better than the examples we heard at the end of the episode?

Jul. 28 2017 03:55 PM
Lydia

The female scientist's explanation elides complete responsibility. should could have said I built this nuke... but that's my job as a scientist. How people use it. Education! Educate them about it.

No, maybe don't use your resources and talents to build something with such obvious nefarious uses. You are responsible for releasing it into the world. It's completely self-centered to say everyone must adjust! Everyone must make themselves safe from this. If people misuse it-- that's not my fault.

In essence she is saying 'I am a computer scientist I have no responsibility'. That is a childish view of ethics.

Jul. 28 2017 03:12 PM
Sir Tafticus from Green Bay, Wisconsin

Another thing that must be considered is the issue of accessibility to NEW or expensive technology. I say this because I cannot afford a 4K television set with 48 inch display, but if I did, your experiment video would be RIDICULOUSLY obvious as a fake. It's not so much about your experiment video, but about how any imperfection in the audio and video is more easily recognizable on a display that is crystal clear, with a speaker that is high definition. So demographics, wealth, and accessibility factor into how easily I can be fooled.

Jul. 28 2017 02:20 PM
Jakob Gowell from Rhinebeck, NY

Once the technology has advanced to generating novel voices not belonging to any particular extant human, Radiolab could use it to generate the audio for the credits without needing to have anybody call in... But what fun would that be?

Jul. 28 2017 01:53 PM
David from San Rafael

The word that everybody danced around but nobody used is "ethics."

Jul. 28 2017 01:33 PM
Greg from Denver

Probably as disconcerting of the technology was the attitude of the researcher helping to develop it. It was plainly obvious that she knew of the ramifications but preferred to handle her cognitive dissonance by stating "she was just a computer scientist".

I also noted the increasingly intertwined relationship between academicians and commercial companies which probably helps to further could the interest of these researchers.

Jul. 28 2017 01:15 PM
Podcast Junkie from AlbuKwerky, NM

Well thanks for that horrifying story.

Science and tech research is like Pandora's box. Once the lid is open, there is no turning back. My naive hope is that this will force people to stop being mentally lazy and actually consider the source and question the intent of the distributor.

Jul. 28 2017 12:40 PM
Aaron from St. George, Utah

WRT my prior comment--please forgive the voice-to-text errors or fat-finger phone keyboard typos I missed while editing that comment on my phone (i.e. rock => raw). And let me add that video and audio Providence will become more important in the future, and perhaps the phrase,"Show me the [block]chain!" will become the mantra of those desiring proof that a recording is authentic.

-Aaron
St.George, Utah

Jul. 28 2017 12:27 PM
Aaron from St. George, Utah

You may want to look into the intersection of block chain technology with developments in audio and video recording equipment. There are those who are working on creating hardware that will cryptographically sign recorded data as it is captured. When used in real-time, connection to the Internet where the cryptographic signatures can be submitted immediately and included in a public block chain, the signatures + block chain inclusion creates an unforgable timestamp proof of existence. So at a live event, if you have multiple people recording video at various angles, using cryptographically secure digital signatures submitted to a public block chain in real-time, The rock, unedited recorded data becomes a powerful proof of what was actually said or what actually happened at the live event. Imagine if political opponents are using this technology and both are recording the same event. So long as one has access to the raw, unedited, signed data signed and in the block chain, it should be impossible for anyone to create fake variations of that event, without those forgeries being easily recognizable when compared to the authenticated data.

While there are some things that may lead to some degree of worry, there are other rising technologies that may mitigate some of the potential issues.

-Aaron
St. George, Utah

Jul. 28 2017 12:07 PM
Jim G from Omaha

This is another example of movies predicting the future. If you watch the old Arnold Schwarzenegger movie running man this is how his character was setup in the movie.

Jul. 28 2017 11:08 AM
Neo from Atlanta GA

Also.. the use of VR physics lectures by Einstein cracked me up !

Back when ARPANet was being created, apparently someone asked the scientists what use would they see for internet. And the scientists said some thing open exchange of ideas on meaningful ideas like art , science etc.

I bet they never saw "I can has cheese burger" coming. :)

I am willing to bet this technology will have a lot more "fun" ;) applications than VR lectures by Einstein. (because a. I like feinman's lectures better and
b. per my 3 year old, talking tomcat and chipmunk songs are more fun than einstein's lectures.. trust me.. i tried.. !)

Jul. 28 2017 10:06 AM
Neo from Atlanta GA

Here's an analogy :

Imagine an artisan in the time of Michaelangelo. Lets say he creates a pigment with a brilliant shade of blue. If someone asked him what use do you envision for this blue, he might have said something like - "Great artists like Michaelangelo are painting up the ceiling of sistine chapel. Trust me, my shade is gonna be so good, people are gonna talk about it for the next 20 years!"

That did happen, for sure, but, it is very difficult for this artisan to understand the profound implications of this new pigment. He might not have imagined Michaelangelo's work would be talked about for a few hundred years. He cannot have imagined its use by masters of painting. Nor could he have imagined a 3 year old in 21st century using HIS pigment to finger paint a stick figure family at her daycare.

Similarly, understanding consequences of fundamental technology always ends up falling short. We typically ask a VERY narrow user base, with very narrow context to contemplate the extent of future uses.

In this article, we explored the uses of this technology with a few, very-well-qualified-but-still-only few, people out of 7 billion.

Question is
1. what uses could come about if this new tool reaches billions of people, many years down the line?
2. How many and what positive consequences could there be ?
3. How many and what negative consequences could there be?
4. Can we figure out if positives would win out or would negatives win out?

Also.. keep in mind.. we human beings (including me) have a "loss aversion bias". We fear a minor loss more than we a wish for a major win. In this context, even if this technology has a positive consequence, we are wired to fear the downsides more?

So - should the fear of unknown downside(s) out weight exploring and creating new technologies (or pigments? )

May be we should let this play out ?

Jul. 28 2017 09:59 AM
Youssef

Oh PLUS the audio replacements of Obama's voice were just so freaking obvious. You could hear different audio quality on the replacements.

Jul. 28 2017 09:52 AM
Youssef

I just watched the Obama video recreation and honestly it's so shitty (sorry Simon) I don't know why you guys are freaking out about it.

What was more convincing was the face2face facial reenactment examples of Bush & Putin I found online. But this example of Obama screamed of manipulation. His mouth was not moving naturally at all, it was being stretched while open, & it looked like it was being puppeteered and i guarantee you if I saw this video just on youtube while browsing I'm pretty sure I would spot it.

Jul. 28 2017 09:50 AM
Josh Marthy from Albany, NY

The future is now

Jul. 28 2017 09:01 AM
Flip Schrameijer from Amsterdam, THe Netherlands

I totally agree with you this is deeply troubling. I'm amazed and - again - very troubled by the total lack of concern by this woman maker (didn't catch her name). Indeed, as you say , "is the world ready for this"?

What frightens me,for one thing, is the possibility that Trump can now deny anything he said on the campaign trail, such as humiliating the impaired journalist: fake news.

I hope indeed methods are being developed which prove beyond any doubt and quickly to tell fakery from reality.

Jul. 28 2017 08:41 AM

Leave a Comment

Email addresses are required but never displayed.

Supported by

Feeds