Joe Rogan Interviews Steve Jobs

This sounds like if you spliced together tape of Rogan and Jobs with scotch tape. An interesting concept but I never felt like I was listening to a conversation between two people. Jobs sounds weirdly far away, he also was never a &quot;fast talker&quot;. Listen to his interview at MIT where he takes questions. He always stopped for long pauses, stares off into space for a minute. He didn&#x27;t rush into responses and tended to nail timing really well.<p>Rogan sounds more believable but I suspect you&#x27;d get the same results sitting down at Audacity and cutting together an MP3. You just have better quality samples of him talking to people in a podcast setting.<p>It&#x27;s an interesting demo but I&#x27;m not sure why I would want this to exist. It seems like technology which you could only abuse, either through the generation of podcast spam or through the production of fake audio. Podcasts are legion and are already cheap to make and based on open standards.

This is mindblowing. Like this could be real, and I&#x27;m learning stuff from it: <i>There&#x27;s an Indian epic that&#x27;s 10 times as long as ...</i><p>There&#x27;s some audio distortion (sounds like clips cut together, little &quot;notches&quot; in the soundscape) but apart from that, and some weirdness in sensing &quot;the spatial location&quot; where this audio was recorded...the concepts and the dialog are amazing.<p>Some parts are weird...but people can be weird. It you tidied this up, and added the right sounds affects and audio processing to this, without the cue that this is AI generated...holy fuck, I think people would believe it. Particularly if you cut it together as a &quot;highlights reel&quot;. Jobs does sound a bit off tho, a bit thin...there should be enough data on him to do a sparse reconstruction of his voice to a level of accuracy beyond human discernment tho.<p>The thing this got wrong about Job&#x27;s voice cadence, tho is: Jobs speaks a lot more slowly and deliberately, and with a lot more pauses, than here. I suspect the cadence &#x2F; timing is not so emphatically modelled by this AI.<p>I think also they&#x27;re missing some emotional trajectory coherence in both their voices. Like the emotional register of the voice does not sound or transition as naturally, and is less diverse.<p>Incredible PoC. AI folks are the new dark wizards. WTF can they not do? That list is shorter

Hi everyone, cofounder of play.ht (the startup behind this podcast) here. let me know if you have any questions.<p>To give more context, the podcast was totally AI generated, the content itself was generated from a finetuned GPT3 on SteveJobs&#x27; biography, the voices were cloned from few hours of both Joe and Steve voices, even though it was tough to get good content for Steve Jobs. And the podcast artwork was generated by SD.<p>We will be releasing more episodes soon which will be even more mind-blowing!

I&#x27;m amused that right now as I post this, the top comment is &quot;This is terrible and sounds like it was cut with tape&quot; and the next one is &quot;this is amazing and indistinguishable from real life&quot;.<p>I guess we have a ways to go with AI, but it&#x27;s already there for some.

The flow of Jobs&#x27; voice is a little off, but wow this is incredible. But also incredibly concerning. Its cool tech, and I guess nothing can stop the inevitability of new technology, but the ability to replicate people&#x27;s voice, speech patterns, and how they look to a pretty accurate degree is mega sketchy. I guess to future proof yourself, just minimize your audio and visual data online...

Do people understand this isn’t just speech synth? The script itself is AI?<p>The introductory bit is AMAZING. Coyly teasing out the introduction of the guest is something Rogan does.<p>I also assume the laughter bits are overused because of the Rogan transcripts, which probably indicate [Laughter]. You’d need a bigger model to know what’s the appropriate laugh for person to person discussion, not the laugh Steve Jobs gives to an auditorium.<p>But overall, really amazing things get picked up.

I&#x27;m sure everything I&#x27;m about to say has already been said, but I would like to add my voice because it seems important.<p>This podcast was enjoyable to me. I was surprised that I enjoyed it so much and I ended up listening until the end. I really liked to listen to Steve Jobs, even knowing that this was some sort of combination of things he has said and generated content, he was very enlightening to listen to.<p>Both Joe and Steve stayed in character as far as I know, the audio and a few laughs that were misplaced seemed like the only flaws technically.<p>Morally I&#x27;m thinking that this falls short unfortunately, but only because i&#x27;m under the impression that permission was not given to use the likeness of either person. Since this is a relatively new way to achieve something like this particular production I believe that these first uses will have the biggest effect and I can understand why the two people featured were chosen. I hope that one day our society respects the right of people to be portrayed and used as they would wish, but I do understand that as of right now the power to enforce those protections are probably not well developed.<p>All in all I&#x27;m very impressed and a bit disturbed by the potential.

I&#x27;m anti-AI so I&#x27;m biased against this from the start, but c&#x27;mon, let the dead rest in peace. It&#x27;s really bad for the culture, imho, to endlessly repackage the same people, ideas, properties, etc. As much as I don&#x27;t care for Musk, Joe Rogan talking to him provides infinitely more insight into the world of today than a fake Rogan talking to a fake Jobs.

While impressive, I think it is extremely bothersome. Did you get JRE&#x27;s permission or the family of Steve Job&#x27;s permission to do this? Considering how much news&#x2F;social media&#x2F;blogs like to cut and paste one-liners taken out of context, I find it unethical to put words into Joe Rogan&#x27;s mouth like this.<p>I get that you needed someone who everyone recognized for a good demo, but I honestly think using someone who is alive was not a great choice.<p>To remove anyone&#x27;s ability to believe this interview actually happened, some other interviewers could&#x27;ve been used, like Walter Cronkite or David Brinkley.

Am I going crazy? This sounds <i>awful</i>.<p>I&#x27;ve never listened to Joe Rogan - but does he really sound like his voice has passed through a square wave? He has almost no intonation. The AI can&#x27;t even pronounce &quot;Swayze&quot; correctly.<p>Job&#x27;s laugh sounds psychotic and out of place. And his voice, again, sounds robotic.<p>The dialogue is utterly nonsense.<p>What&#x27;s going on with the hype here?

I&#x27;m surprised how real Rogan sounds and how Jobs does not. Why is that?

This has some obvious rough edges but still it&#x27;s a harbinger of things to come. Seems this was trained on recordings of Jobs&#x27; on-stage voice rather than his casual conversation voice. The rough edges in this will be smoothed out over time by new tricks of AI trade. This will get crazy when combined with an AI Jobs that paces around a virtual stage in a black turtleneck hyping AI-invented products.

So we have ai replacing humans; but more, they will even be assuming our names and the internet will be full of fake conversations and articles and fake stories allegedly by these replaced people<p>There needs to be some laws about this asap

The really bizarre thing is that I found myself enjoying this podcast, perhaps even more than the average Joe Rogan podcast. This didn&#x27;t feel quite as hollow as the average AI-generated content.

Can you imagine how much history the next Stalin or Mao will be able to re-write in the post-truth era using a good version of this?<p>This is the foundation of Propaganda 3.0<p>Welcome to the era of <i>Psy-Weapons of Mass Madness</i>.

Putting the ML conversation to one side, this is a fantastic piece of marketing.<p>It&#x27;s perfect for getting attention from a wide audience and is a great demo of their product.

This really feels like next level stuff--as Jobs would love to note, that&#x27;s the magic of integration. But, it&#x27;s really not novel--just someone putting together the building blocks of foundational ML models already out there.<p>I&#x27;m really excited to see version 2 and so on of this.

Wow Rogan sounds very realistic. Is the voice completely generated or are some words&#x2F;phrases &quot;borrowed&quot; from real sample data portions?

What I&#x27;m waiting for is when these imagined conversations are generated for an audience of <i>just me</i>. When references I understand and abstractions I&#x27;m familiar with are tossed around with ease, but unfamiliar concepts are expanded on and picked apart, as you might with a good friend.<p>If the speakers are brilliant I often don&#x27;t want to actively be part of a conversation, I just want a personal performance. Though I imagine in 10 years, with a tricorder on my face (brain sensor included), I&#x27;ll probably always be part of the conversation&#x2F;performance, however passive.

What is the value proposition for such a thing?<p>For people complaining that it is not good enough, it may sound a bit off because its ML and also depends on the availability of good data. Over time, the algorithm will fix these issues, may be.<p>For me the overarching thing is, why should this exist?<p>1. Are they auditioning for a deep fake service and picked the most eye catching, legally controversial thing possible?<p>2. Do they think it has a viral potential and so it could be a continuous source of revenue?<p>Either way, for me it&#x27;s a PoC that shows how trivial it is to fake someone&#x27;s voice and think about the ramifications of this for our society.

Very engaging after the first act. Imaginary Conversations is already an established sub genre of history books and I can&#x27;t wait to see where this style of hybrid curation &#x2F; creation goes, surely into the present like this. Maybe a.i. voice debates will be so offensive to the popular pundits being cloned it will encourage more real live debates again, giving people exactly what they want just to compete with underground a.i.<p>I expect soon there will be Unauthorized Biographies written in the first person with the audiobook read by the subject&#x27;s a.i. voice. What a trip.

It sounds good on a first glance, but then the topic switches are totally absurd.<p>Like Steve talking about LSD all of a sudden, then Joe Rogan asks about Newton (which Jobs didn’t make) and Steve starts talking about… product and iteration?<p>It sounds like someone is superficially trying to act like Jobs would act, which … it kind of is!<p>I needed to turn it off after 10 minutes as it obviously goes nowhere. But really interesting experiment nonetheless, and better than I would expect.

I wonder if people on their deathbeds will start recording themselves so that AI can immortalize them for their loved ones after they&#x27;re gone.

The opening minute or two sounded a little off, but either Jobs&#x27; cadence sped up or I got used to it, and it got it better as it went on. The last few minutes were astounding - Jobs riffing about the good and bad potential of technology, and the final bit about throwing computers - if that was not either heavily curated or copied from pre-existing material I am quite impressed.

Is this scripted by an AI or scripted by a human and simply voiced by an AI?

This is a misleading title since it&#x27;s a fake.

I too throw my MacBook out the window every now and then. Not to check if it still works, but to get it to work when it stops working

Mind is blown.. it&#x27;s verry good. Although there&#x27;s no real coherence in the &#x27;ainterview&#x27;, it&#x27;s very convincing. In terms of TTS, there are some minor glitches here and there, but overall I&#x27;m super impressed.<p>one of the things was the &quot;sell you a quarter of a car&quot; part. I couldn&#x27;t find a google hit for that.

Steve jobs is spinning so fast in his grave that Apple will no doubt add generators. This should make them carbon neutral.

&quot;People will accept your ideas much more readily if you tell them that Benjamin Franklin said it first&quot; can now be amended to &quot;people will be more willing to listen to your ideas if you synthesize a podcast of Joe Rogan and Steve Jobs talking about it.&quot;<p>I don&#x27;t think anything being said is new wisdom or truth. We&#x27;ve heard these same things said elsewhere. However, because we spliced together the voices of Joe and Steve, you&#x27;ll get a different level of engagement from a different set of communities. It&#x27;s already hard enough to separate form and substance on the internet. I&#x27;m not sure what this adds.

The 1st thing I noticed is that Jobs voice sounds like he&#x27;s giving a talk to a large audience as opposed to a one-on-one conversation. Is it because most of his clips are him doing a monologue?

Okay, this is moderately terrifying.<p>Really though, who would&#x27;ve thought we&#x27;d invent artificial creativity before artificial general intelligence?<p>I mean when you see things like this, how far can general intelligence be?

Something often discussed is how things like this can be used to make incriminating fake audio of other people&#x27;s voices. Seems obvious that it wouldn&#x27;t be too hard to find signs of that in the digital files.<p>My question is has anyone made a filter that goes the opposite direction, ie makes your voice on phone or zoom calls sound like it&#x27;s been AI generated by inserting specific artifacts?<p>We might have an interesting arms race ahead of us. Maybe a good time to get into digital forensics.

Uncanny Valley meets bad podcast editing. Sounds like somebody cut+pasted a bunch of Steve&#x27;s Ted Talks &#x2F; Apple promos. The laugh was really dissonant.

This was very uncomfortable for me to listen to. I felt compelled to close out of the tab 3 minutes into it. While I can appreciate that the AI has stitched together dialogue read by synthesized voices, ultimately it sounded like nonsense and an invasion of privacy for both parties. Maybe I&#x27;m just weird and have a low tolerance for that uncanny valley feeling.

My expectations weren&#x27;t that high to begin with but... hmmm... What should I have been expecting in context of what the AI is actually doing? Is it automatically cutting together clips to fit a script? The cadence of Jobs just seems strange and comes off as being stitched together. Surely an AI that can get this far could add some natural gaps to the conversation.

How many bits of information was there in the prompt. If it is was a very small amount, this is mind blowing!<p>Being trained off Steve&#x27;s autobiography dampens how impressive this is, but the grammatical structure and flow of sentences is beyond belief. We did see this in GPT3. But the responses seem more coherent, probably because of the source material- Jobs biography!

This is both impressive and unbearably, unlistenably annoying. I had to stop like 45 seconds in.<p>It sounds exactly like Rogan interviewing Jobs as generated by an AI from 2022. Definitely nailed that. Ten thousand percent.<p>In a remarkable milestone, AI can now match the intelligence of not one, but two, mindlessly rambling people with dementia.

I wonder how much more effective this would be if the prompt wasn&#x27;t &quot;this was generated by AI.&quot; It would be really interesting if someone with a popular podcast injected an AI conversation into their feed and didn&#x27;t tell their audience for a couple weeks. A real life Turing Test.

A thought this prompted: I absolutely cannot wait to see how litigation erupts as this stuff improves. Digital forensic evidence investigation is going to become a massive industry.<p>People are going to be imprisoned and put to death for things that never even happened. Unreal times we&#x27;re living in.

Why not just listen to an actual interview of Steve Jobs by Terry Gross? <a href="https:&#x2F;&#x2F;freshairarchive.org&#x2F;segments&#x2F;steve-jobs-future-web" rel="nofollow">https:&#x2F;&#x2F;freshairarchive.org&#x2F;segments&#x2F;steve-jobs-future-web</a>

Joe is absolutely on point.<p>Jobs, not so much. I think Jobs was trained from public presentation or something.

Product idea: audio book but with the (synthetized) voice of their author. Does this exist?

Entirely generated by artificial intelligence? So this text was generated by GPT3?

amazing&#x2F;terrifying etc.....<p>It&#x27;s interesting many people&#x27;s reactions to that of Jobs are &quot;he wasn&#x27;t like that&quot;, &quot;doesn&#x27;t sound what he would say&quot; etc..... (1) a dead person, freaky, (2) countering people&#x27;s biased &quot;memory&quot; of someone from their past.... what we pick and choose to remember... going up against something generated (with it&#x27;s own biases&#x2F;problems) from the content that person produced over years and years...what seemingly should be a really decent representation of their -isms&#x2F;personality.

It&#x27;s a cool demonstration. I was doing a lot of GPT-3 stuff last year, got terrified and felt hopeless in face of where AI was going.<p><a href="https:&#x2F;&#x2F;semiosis.github.io&#x2F;about&#x2F;" rel="nofollow">https:&#x2F;&#x2F;semiosis.github.io&#x2F;about&#x2F;</a> <a href="https:&#x2F;&#x2F;semiosis.github.io&#x2F;posts&#x2F;imaginary-internet-survival-guide&#x2F;" rel="nofollow">https:&#x2F;&#x2F;semiosis.github.io&#x2F;posts&#x2F;imaginary-internet-survival...</a><p>But I put it all down to do philosophy, then some soul searching and God turned up. Spirit, astral projection, lightsoul, God, Christ, heaven -- all real.<p><a href="https:&#x2F;&#x2F;semiosis.github.io&#x2F;posts&#x2F;the-pathway-to-gnosis&#x2F;" rel="nofollow">https:&#x2F;&#x2F;semiosis.github.io&#x2F;posts&#x2F;the-pathway-to-gnosis&#x2F;</a>

Most mindblowing thing I&#x27;ve seen in recent memory. With the amount of available training data, isn&#x27;t it now just a small step to AI-generate new episodes of existing podcasts? Exciting and scary.

Fantastic marketing execution. Most surprising was the gravitation of the conversation. I&#x27;m really curious about the script behind the voices. Was it written by a person or ai as well?

This is so good that I can&#x27;t tell if it&#x27;s AI generated or a bunch of clips of Joe Rogan and Steve Jobs stitched together decently well.<p>The &quot;belief&quot; threshold has been crossed.

&gt; it would seem contrived and everyone would see through it<p>It&#x27;s as if he&#x27;s describing this very podcast. Nevertheless I enjoyed listening to this as much as any other JR podcast.

Obviously it has some kinks that need to be worked out and it sounds spliced, and Joe never starts the show “Hello to all you freak bitches” but was better than I expected.

A lot of people aren&#x27;t going to like this. The general public perception is going to be that not only are AI after our creative jobs but even our identities.

Similar project: <a href="https:&#x2F;&#x2F;lexman.rocks" rel="nofollow">https:&#x2F;&#x2F;lexman.rocks</a>

Scary. Not sure how democracy survives when adversaries can make people say anything they want them to.

Does anyone else find all the little AI-powered non sequiturs and nonsensical points as funny as I do?

Only the voice is AI right? Is the whole conversation flow generated by AI too? That&#x27;d be crazy.

&gt; 01:04 &quot;And I can&#x27;t even say his name&quot;<p>Well, it looks like a big fail right off the bat.

Ok, I&#x27;m ready for the Max Headroom podcast to be a permanent fixture of my life

What’s the process here- generate a transcript with GPT-3, deepfake the voices?

Alan Watts next, please. :)

Another podcast has already been doing deepfake celebrity interviews with ai-written scripts (but with a humorous tone) for a few months:<p><a href="https:&#x2F;&#x2F;m.soundcloud.com&#x2F;jn2022" rel="nofollow">https:&#x2F;&#x2F;m.soundcloud.com&#x2F;jn2022</a><p>They&#x27;re all hosted by a fake Lex Fridman. The Elon Musk ones sound pretty convincing.

I like how Jobs is just using his WWDC voice the entire time

This is the best podcast I’ve ever heard.<p>Fantastic job. Excellent work.

Reminds me of Neuromancer&#x27;s Flatline construct.

This is incredible!

Joe Rogan is the best interviewer of our time.

Is the content AI generated as well?

Does the AI spread covid misinformation too?

The first 20 secs of dialog from Jobs is not believable. He never did chit-chat. He wouldn&#x27;t give a toot about some podcast. His mission would be to get to the point - himself, his company, or a big idea that intersects one or the other. He wouldn&#x27;t try to flatter the the host.<p>Bro Jogan, on the other hand, sounds very believable.<p>This might have something to do with the ratio of Bro&#x2F;Jobs training material.<p>Still, this technology has huge potential.