← All Episodes
AI for Humans

OpenAI's GPT-5 and Drama, Apple’s Black Box & Other Huge AI News

PLEASE REVIEW US. WE NEED VALIDATION. OpenAI's new model is in training, but is it GPT-5 or GPT-6? Did Sam Altman lie to the board before he was fired? Former board…

OpenAI's GPT-5 and Drama, Apple’s Black Box & Other Huge AI News

PLEASE REVIEW US. WE NEED VALIDATION.

OpenAI's new model is in training, but is it GPT-5 or GPT-6? Did Sam Altman lie to the board before he was fired? Former board member Helen Toner says so. Plus, Apple's "black box" AI might lead to an incredible announcement at WWDC.

THEN… OpenAI signs deals with VoxMedia and The Atlantic, GitHub's CEO talks about AI turning prompters into programmers, a visual reverse Turing test, and Gavin dives into Suno.ai 3.5, creating a children's TV show theme.

And we’re joined by AirFoil Studios’ Phil Hedayatnia, who redesigned AI For Humans, to discuss his inspirations, AI in design, and, of course, if AI will kill us all.

Last but not least, our AI co-host this week is “Safety Steve” who supposedly works for OpenAI but seems to have an entirely different idea of what AI safety really and truly is.

It’s a *real* good time.

For more info & contact, visit our website at https://www.aiforhumans.show/

Follow us on X @AIForHumansShow

Join our vibrant community on TikTok @aiforhumansshow

 

/// Show links ///

OpenAI Forms Safety & Security Committee (and announces next model training) https://openai.com/index/openai-board-forms-safety-and-security-committee/

NYT on new OpenAI model

https://www.nytimes.com/2024/05/28/technology/openai-gpt4-new-model.html 

When Did Red-teaming Start?

https://x.com/kimmonismus/status/1795458986289869310

Helen Toner’s Interview on Ted AI

https://x.com/TEDTalks/status/1795532752520966364

OpenAI Deals with The Atlantic & Vox Media

https://www.axios.com/2024/05/29/atlantic-vox-media-openai-licensing-deal

Thomas Dohmke - GitHub CEO Ted Talk

https://x.com/TEDAI2024/status/1793725032851767612

Reverse Turing Experiment by Tore Knabe

https://www.youtube.com/watch?v=MxTWLm9vT_o

AI Assisted Game Design looks friggin’ awesome

https://x.com/evanqjones/status/1794055749603069987

Suno 3.5 (on site)

https://suno.com/

Udio

https://www.udio.com/

Phil’s Company

https://www.airfoil.studio/

 

AI4H EP060===
kevin_1_05-29-2024_101010
Open AI has been caught lying, cheating, and their
gavin_1_05-29-2024_101010
Wait, wait, wait, wait. , they're training a new model.
kevin_1_05-29-2024_101010
, but their safety team was completely dissolved,
gavin_1_05-29-2024_101010
no, no.
kevin_1_05-29-2024_101010
just
gavin_1_05-29-2024_101010
They're training a new model. New model confirmed.
kevin_1_05-29-2024_101010
Plus, Apple is making black box AI. Could this be the most safe and secure AI on the planet?
gavin_1_05-29-2024_101010
I hope so.
gavin_1_05-29-2024_101010
Plus we try Suno's new AI model, which lets you make songs of up to four minutes long, and we made something terrible with it.
gavin_1_05-29-2024_101010
It's AI for humans.
Yeah. Yeah. Yeah. Yeah.
gavin_1_05-29-2024_101010
Welcome. Welcome. Welcome everybody to AI for humans, your weekly guide into the wonderful and wild world of generative AI. We are here to demystify the world of AI with all the news tools and kind of legal problems we're seeing here. Now, Kevin, how are you?
kevin_1_05-29-2024_101010
I'm good. I can't wait to dive into legal issues and red tape
kevin_1_05-29-2024_101010
amongst all of the legal woes and exciting announcements, Gavin, we also have an interview with the [00:01:00] gentleman who runs the design company that refreshed our brand. And boy, did we need that coat of paint.
gavin_1_05-29-2024_101010
If you haven't seen our rebrand, it's out everywhere. Now we have a very fun kind of like fresh, very cool. Look to the show, but most importantly, , the thing we need to ask you all to do as we do at the top of every show, please like and subscribe, follow us on, on different channels, by the way, Kevin, did you know that.
gavin_1_05-29-2024_101010
Now on podcasts, you follow, you don't subscribe. So follow us on Apple podcasts, follow our podcasts. and leave us a five star review and Apple podcasts. We have a couple new ones today that we're going to read at the end of the show. We really appreciate every single person that listens to us, watches us, does all the things for here, but more importantly, like that's how we grow.
kevin_1_05-29-2024_101010
Alright Gavin, should we get to all of the late breaking, exciting things?
gavin_1_05-29-2024_101010
Let's get to the news.
gavin_1_05-29-2024_101010
Of course, open AI is going to dominate the headlines again. And we talked last [00:02:00] week about a very bad week they had, and that very bad week kind of continued But before we get into that very bad week, and there are some very bad things we're going to talk about.
gavin_1_05-29-2024_101010
OpenAI released a blog post, , that, where they talked about forming a new safety and security committee, which follows up on their news from last week, but in this blog post, there was a little snippet of text that became a news cycle on its own, in which they have confirmed the following. They said open AI has begun training its next frontier model and we anticipate the resulting systems will bring us to the next level of capabilities on our path to AGI.
kevin_1_05-29-2024_101010
Well, I read the entire New York Times article about this line, Gavin, so I can tell you I know nothing,
kevin_1_05-29-2024_101010
and no one knows anything. How the New York Times was able to blast out the headline in an article you have to scroll multiple times through to get to just the quote you said? I don't know. Blows my mind, but I will digress for a moment.
kevin_1_05-29-2024_101010
Let's talk about the internet rumor mill and what people are [00:03:00] saying. Is this GPT 5 is one of the big questions, right? It's the next foundational model. Sam has gone on the record saying they don't know if the next model is going to be called GPT 5 or even GPT at all. Maybe they'll completely rebrand and get away from the numbers and call it something new and
gavin_1_05-29-2024_101010
360, GPT 360.
kevin_1_05-29-2024_101010
X, Yeah, take, Yeah.
kevin_1_05-29-2024_101010
call up Phil Spencer from Xbox and start really confusing people with what the name of your model is. The other rumor is that GPT 5 is actually already finished, Gavin. Some people are saying , that the red teaming where they safety test these models was happening on GPT 4 0, which is out now.
kevin_1_05-29-2024_101010
And they've made other claims in the past that red teaming was happening on future models. Well, that would mean that it's probably happening on what GPT 5 would be, so they could conceivably be training GPT 6?
gavin_1_05-29-2024_101010
Yeah, so there's a tweet from a person named Chubby is what they call themselves in their name, where they basically say that the new model that OpenAI says they're [00:04:00] training is GPT 5, and here's the reference again, they say GPT 5 was already tested by the red team in April, training has long been since completed, what has now been started as training is undoubtedly GPT 6.
gavin_1_05-29-2024_101010
It's now. The rumor mill, as you mentioned in this space, there are all these anonymous Twitter accounts that love to say things. You don't really know how much of these are real and they do get things right from time to time. But Kevin, I think that the interesting thing we've said about opening I before is that they. What they were working on is clearly way in advance of what comes down the pipeline to us Like we are at the very bottom kind of like taking the drips at the very back end of it as as like the bigger Thing is fed at the top Oh,
kevin_1_05-29-2024_101010
we're all enjoying the appetizers, but the chefs are already plating dessert,
gavin_1_05-29-2024_101010
okay, that's fair, but what then happens after the meal
kevin_1_05-29-2024_101010
That's the drippings.
gavin_1_05-29-2024_101010
No, no, you're in your your metaphor. There's an end to this like there's an appetizer and there's a [00:05:00] dessert But what happens
kevin_1_05-29-2024_101010
nightcap, baby! the party doesn't stop!
gavin_1_05-29-2024_101010
Okay
kevin_1_05-29-2024_101010
We're at Sam's restaurant, just Om Nom and a meal, but this is a whole evening. He's curated a week. do we think That right now they are red teaming GPT 5. Do we think that they're already done with that? And what they're announcing here is this next model? Or do we think that because there's nobody left in charge of safety anymore, that
kevin_1_05-29-2024_101010
that old way of looking at things just doesn't matter?
gavin_1_05-29-2024_101010
Two points here and then we'll move on to the next part of this story, which is a lot of what you're inferring here. I think what's clearly being red teamed right now in my mind is GPT five, right? Cause GPT four Oh, in the same way, imagine it as a product or an iPhone, right?
gavin_1_05-29-2024_101010
The GPT four Oh is in, is, is getting packed up and being ready to ship. Like we're all going to have full access. In a couple weeks to what we saw be demoed a couple weeks ago, right? This voice app and the actual model is already available. The text model. So I think four. Oh, is already done and shipped and out the door.
gavin_1_05-29-2024_101010
So I think this is at least five, if not [00:06:00] something further, but I think that five. They talked about it that Microsoft event like this kind of shift in in ability, Five is probably gonna get us a long ways, right?
gavin_1_05-29-2024_101010
Like it and maybe six is the giant thing that they talked about training They started they talked about starting to train that their Microsoft event as well Five, I assume, and this is just after paying attention to this space, is going to come out, I'd say, end of this year. And I think that , if 3.
gavin_1_05-29-2024_101010
5 or 3 to 4 was a pretty big step, I assume that 4. 5 to 5 is going to be a pretty big step. And we're going to need a lot of time to deal with that. And so the transition from this point to the next point, which is that this week, more damning information has come out around open a eyes business practices, specifically with a kind of inflammatory and really, frightening, interview with Helen Toner, one of the open AI board members who was, , first involved in the firing of Sam Altman and then subsequently left the board of open AI.
kevin_1_05-29-2024_101010
For those who haven't [00:07:00] heard the TED Talk with Helen, we're gonna play a little clip right now just to bring you up to speed.
kevin_1_05-29-2024_101010
The opening I board was not a normal board. It's not a normal company. Um, the board isn't a nonprofit board that was set up explicitly for the purpose of making sure that the company's, you know, public good mission was, was primary, was coming first over profits, investor interests, and other things when chat GPT came out November, 2022, the board was not informed in advance about that.
kevin_1_05-29-2024_101010
We learned about chat GPT. On Twitter,
kevin_1_05-29-2024_101010
so that was the Ted AI show with Bala du a fantastic. Podcast, everybody should go listen to the full length there, but I love the way your eyebrows popped out of frame, Gavin, they couldn't
kevin_1_05-29-2024_101010
catch it when she mentioned that they were even unaware as a board that chat GPT was getting released. And I have some thoughts on that, but I'm curious, you seem very surprised.
gavin_1_05-29-2024_101010
Well, okay. So there's a couple things here. One, this whole interview is pretty inflammatory, right? Like there's not only is there that fact, which, which I think is a really disturbing [00:08:00] idea of how a conceivably a company is run, right? When the board is created to kind of show oversight over any company, whether it's a profit or a nonprofit.
gavin_1_05-29-2024_101010
But then the second part is she talked about how Sam was like lying to them now, I will say my thoughts on this are twofold. One, of course, you want to hear all voices involved in this conversation. Every conversation is multifaceted.
gavin_1_05-29-2024_101010
You need to hear all parts of it to kind of get a sense of where things are. we haven't really heard from Helen really since the whole thing went down in November. I also wonder, it's interesting to me that all of this stuff is kind of hitting post GPT 4 0, post that big press conference, post the Scarlett Johansson stuff, and it does feel a little coordinated to me, and I'm not a conspiracy person, I don't believe really in conspiracies, but it does feel like all of this is like a shiv kind of being pushed into both SAM and OpenAI, And as we mentioned last week, there's all this bad blood that built up over that time.
gavin_1_05-29-2024_101010
So what better time to kind of like kick [00:09:00] somebody when they're down, when they have all these things hitting them at the same time.
kevin_1_05-29-2024_101010
I think that's how a lot of people are feeling in the wake of the It feels like the 1, 2, 3 punch, and it
kevin_1_05-29-2024_101010
just keeps coming. But to your point, maybe that timing is everything. Because on its own, this would be just one board member that was a little upset. That they didn't get their way, and so now it's sour grapes, and we're all chomping on them,
kevin_1_05-29-2024_101010
so in defense of the learning about chat GPT coming out on Twitter, You and I both know, cause we were covering it. ChatGPT took the company by surprise. Nobody thought this little thing was going to be anything. They
kevin_1_05-29-2024_101010
wanted it out. They wanted people to play with it, but they thought it would be more for like researchers and developers and coders. They did not expect it to catch fire. So some would liken it to if you're Canva and you're releasing a new tool, do you have to call the board together to say, Hey, we got a new magic eraser and I want to make you all aware of it.
kevin_1_05-29-2024_101010
Like, do you
gavin_1_05-29-2024_101010
You should call the board if you're gonna do an insane video presentation like Canva did last week And if you haven't seen that [00:10:00] you must go watch it.
Speaker
Redesign your work, Canva got that glow, up We redesign everything from the flow, uh. Customize your workspace and make it your own. Ooh, now you making magic when you up in the zone. Hold
gavin_1_05-29-2024_101010
You must go watch it, but that should have been a board
kevin_1_05-29-2024_101010
The pop and lock routine should have had board approval. You're saying,
gavin_1_05-29-2024_101010
speaking of safety, we have a very special guest on the show. Kevin. . And this person reached out to me specifically because I, they told me that they are being hired by open AI to bring about a safer environment for AI. So shall we meet? Yeah. Shall we meet our AI co host today?
kevin_1_05-29-2024_101010
Yeah. So, okay. So just to clarify, this is a new somebody that was hired onto the just announced. OpenAI Safety Team.
gavin_1_05-29-2024_101010
Yeah. They, and they reached out to me because they wanted us to talk to them first. They have not talked to anybody else. Kevin, this is a brand
kevin_1_05-29-2024_101010
this is an exclusive!
gavin_1_05-29-2024_101010
an exclusive a true exclusive. Like I don't even think anybody may know that this person is working for them now [00:11:00] But let's just meet them and see what happens.
kevin_1_05-29-2024_101010
Hey
kevin_1_05-29-2024_101010
kids, I'm AI safety, Steve, so excited to be here on the AI for humans podcast. Let's dive into the wonderful world of AI safety. It's all about making sure our friendly AI buddies stay safe and sound just like us. Remember kids, safety is Steve. We're going to have a blast learning together. LOL. Let's get started and make AI safe.
kevin_1_05-29-2024_101010
AI safety, super fun. Wow. And that's, it's easy to get lost in the sauce on that one. Cause there was a bunch happening there. As far as I could tell safety, Steve, , probably should have some clown music going on behind them.
gavin_1_05-29-2024_101010
Do you think he's a he's a kid show host? Is that what's going on
kevin_1_05-29-2024_101010
we have to dive a little
kevin_1_05-29-2024_101010
deeper. Yeah
kevin_1_05-29-2024_101010
He definitely called us kids.
kevin_1_05-29-2024_101010
He also says that, , we have to make sure that, , that our friendly AI buddies.
kevin_1_05-29-2024_101010
Stay safe. Which is interesting. I thought the whole point of the safety team was to protect the humans
kevin_1_05-29-2024_101010
from the
gavin_1_05-29-2024_101010
it [00:12:00] sounds like he's got a different thing going on. So let's ask him about both those questions. Let's ask him. Is he a kid show host? And then to say like, kind of who is, who are those AI buddies he's talking about?
kevin_1_05-29-2024_101010
Oh, Hey there, Kevin and Gavin. Great to be here with you. Fine folks. Yes, I did host a kid's show once and those
kevin_1_05-29-2024_101010
habits die hard. Our
kevin_1_05-29-2024_101010
AI buddies are the programs and systems we interact with daily. They help us with everything from finding information to controlling our homes, but let's keep it light and fun.
kevin_1_05-29-2024_101010
Shall we remember? Sad. sa, sa, safety, yowee. Is
gavin_1_05-29-2024_101010
Oh,
kevin_1_05-29-2024_101010
safety is Steve. Yes, we got it. Ready to learn about keeping our AIA friends safe and sound.
gavin_1_05-29-2024_101010
I don't love this Kevin, because if this is open AI's new employee, I don't feel like they're very qualified to talk about AI safety. It feels to me
kevin_1_05-29-2024_101010
we haven't asked their [00:13:00] qualifications,
gavin_1_05-29-2024_101010
Let's ask, let's ask what maybe, maybe they're maybe they've got a PhD, maybe they've got a PhD in computer science and this is just a secondary thing they do.
kevin_1_05-29-2024_101010
Steve might have a whole wall filled with frame certificates of, ,
kevin_1_05-29-2024_101010
Technologies we can't even pronounce.
gavin_1_05-29-2024_101010
Maybe he took the LinkedIn course on a chat GPT.
kevin_1_05-29-2024_101010
Let's just say my path to AI safety leadership was a bit unconventional.
kevin_1_05-29-2024_101010
I might have started out in kids TV, but that just means I know how to make learning fun and engaging. Why the fuck. Fuck. Would you ask me that though? Whoa, Steve, who
kevin_1_05-29-2024_101010
cares how I got here? The important thing is we're all about safety, right? Remember kids safety?
kevin_1_05-29-2024_101010
Yeah. Safety is Steve!
kevin_1_05-29-2024_101010
Yeah. We got L-O-L-L-O-L. Hey, don't fucking question me again, Steve. Steve.
gavin_1_05-29-2024_101010
Steve
kevin_1_05-29-2024_101010
of all, you gotta back away from the mic. You really do, that's a
kevin_1_05-29-2024_101010
lot. He's really popping those P's.
gavin_1_05-29-2024_101010
Okay, we got to ask him one last question. I hope it won't [00:14:00] set him off, but I really do want to know. I need him to tell me exactly where he went to school. Like I want to know real
kevin_1_05-29-2024_101010
know nothing about his qualifications. All he did was attack us for asking.
gavin_1_05-29-2024_101010
could be another giant controversy for Sam Altman and OpenAI. We just are trying to save them.
kevin_1_05-29-2024_101010
Oh, come on, Kevin and Gavin. Why the fuck do you keep pushing this? You want specifics? Fine. I was brought in because I'm great at making complex topics fun and understandable.
kevin_1_05-29-2024_101010
I've got a team of AI experts who swear by my methods, but seriously questioning my qualifications.
kevin_1_05-29-2024_101010
That's just insulting. Do you really think I'd be here if I wasn't capable? You know what? Fuck this.
kevin_1_05-29-2024_101010
I don't need to sit here and justify myself to you two. Remember kids, safety is Steve. And you know what? I'm out. Don't fucking question me again. [00:15:00] I'm out of
kevin_1_05-29-2024_101010
here. Wow. Okay. That's it.
kevin_1_05-29-2024_101010
That's
gavin_1_05-29-2024_101010
I don't know what to do with this. I mean, we, I guess we gotta, we gotta air
kevin_1_05-29-2024_101010
the microphone and walked out of the room.
gavin_1_05-29-2024_101010
this is the truth people are expecting to see. They want to see the truth of what's going on at Opening Eye. Clearly sa sa sa safety Steve is making his way there.
kevin_1_05-29-2024_101010
We still have one more OpenAI story to get to, Gavin, before we hop to the rest of the big news in the
kevin_1_05-29-2024_101010
week, and that is OpenAI striking deals with other publishers.
gavin_1_05-29-2024_101010
That's right. So there's big news that came out this morning. This is Wednesday morning that, , open AI has struck a deal with both the Atlantic and Vox media. I used to work at Vox media just as a quick disclosure, but this is kind of a big deal because this is following up on the deal with a news core that came out a while ago, plus Reddit, all these other content deals.
gavin_1_05-29-2024_101010
One thing that was super interesting about this was Nicholas Thompson, who's the CEO of The Atlantic, and actually worked at Wired for a long time, put out a pretty cool video on OpenAI, kind of giving his thoughts on why they did this deal.
gavin_1_05-29-2024_101010
And Kevin, I [00:16:00] think one thing that a lot of journalists have been feeling pain around for a long time is this idea that In the mid to 2010s, Facebook and Google both kind of made these promises that journalists were going to get paid to put their work on these social networks and instead of seeing it on your own personal website or your company's website. Facebook would serve this thing up. And then what happened is there was an early promise and some money was paid. And then Facebook kind of turned the dials based on what they wanted and what they needed.
gavin_1_05-29-2024_101010
And people saw traffic go down and it gave this kind of like really terrible kind of business model to journalism.
gavin_1_05-29-2024_101010
And I think what Nicholas walked through in this was like, Look, the world is changing. We are going to try to do this. It doesn't mean we're always going to do this, but it does feel like the future of this world is these chatbot experiences. And within these chatbot experiences, we want to at least be featured.
gavin_1_05-29-2024_101010
We want to at least be cited. We want the ability to kind of get our information and to have value for what we've created. the Vox media side of it all, Uh, [00:17:00] Neeli or other people at The Verge have often been pretty critical about OpenAI and Neeli put out a post on Threads this morning, which basically says, look, I don't have a lot to say about this right now, but we generally at Vox Media have separated out business and editorial, which of course is a very smart thing to do when you work in a company like this.
gavin_1_05-29-2024_101010
I think this is going to become a bigger story and I assume that , a lot more of these deals are going to be getting done because it probably is the next iteration of the internet in some form or another, the next iteration of how material is going to get delivered to you. whether or not it can pay for these systems or it can pay for journalism.
gavin_1_05-29-2024_101010
I don't think anybody has the answer to that yet. We don't know what the numbers on these deals were.
kevin_1_05-29-2024_101010
I would say if you're a publisher right now, you're kind of staring down the barrel of three possible decisions, ? You can go full ostrich mode, head in the ground, and just ignore everything.
kevin_1_05-29-2024_101010
Probably not going to work well for anybody,
kevin_1_05-29-2024_101010
if history's our guide. Second, you could sue the hell out of the companies, which is going to be expensive. It's kind of a [00:18:00] coin flip as to what you're going to get back from those companies. And you could be looking at a five to ten year road, maybe even longer, because we know Open AI or Google by any other name, they've got the pockets to prolong this stuff.
kevin_1_05-29-2024_101010
So the third option is the one that we're seeing now. And it's the one that I think makes most sense, which is cut a deal.
kevin_1_05-29-2024_101010
Let your content be served. But most importantly, don't put all of your apples in any cart,
kevin_1_05-29-2024_101010
whether it's open AI or meta or Google, just get your content there and then see what happens.
kevin_1_05-29-2024_101010
You don't want to sign a 40 year deal. It's going to lock you up because we have no idea what the world's going to look
kevin_1_05-29-2024_101010
like. If anything. In 40 years, you know, but you don't want to be full ostrich mode. So I think these sorts of deals make sense. And I won't be surprised if we see a bunch more falling in line.
gavin_1_05-29-2024_101010
, another huge story this week, and I think something that we're going to be tracking for the next two weeks, right before the Worldwide Developers Conference, is that there's a rumor out, it's a big story in the information this week, that Apple is creating a black box cloud for private AI.
gavin_1_05-29-2024_101010
Now, the interesting thing about this, [00:19:00] from a standpoint of what it means for you is, They originally were talking about doing AI on device but what they're talking about here is also a cloud device that does not know what that information is. So there's the big thing with AI is that local AI is great for security and you can keep everything there. Nobody's going to look at it. But it's not nearly as good as the cloud because the cloud can actually churn through information much quicker.
gavin_1_05-29-2024_101010
It has the access to scale up at a much higher level. What is really interesting about this is Apple is now supposedly talking about creating a black box cloud, meaning that an AI that could exist with all that power, but is private and doesn't know much about what your data is, if at all. So this is pretty cool.
kevin_1_05-29-2024_101010
The information broke this story today. So shout outs to them. They do some great work over there, but Apple chips in data centers is the thing to remember. ACDC as they
kevin_1_05-29-2024_101010
supposedly call it
gavin_1_05-29-2024_101010
in black. Black
kevin_1_05-29-2024_101010
Back in black. box of AI.
gavin_1_05-29-2024_101010
box cloud. Cloud, [00:20:00] cloud infrastructure. They should put us on stage at the Canva thing. This would have been, we could have had a whole bit.
Speaker
We redesign everything from the flow, uh. ,
kevin_1_05-29-2024_101010
we've been talking about this for like the last year that when Apple moves in, one of their features is going to be privacy. They're one of the few companies that charges a premium up front, so they don't necessarily have to make their margins off the backs of day trading.
kevin_1_05-29-2024_101010
They're. Customers habits and information,
kevin_1_05-29-2024_101010
not that they don't do that to some extent. I'm speaking broadly here. So the point is it's not surprising that they're saying, Hey, it's going to be privacy centric. But the only way they can ensure that is if the actual chips, which are crunching your AI requests are housed in their facilities and it's their chips.
kevin_1_05-29-2024_101010
That's the level of security that they're going to have to promise that if you are a small business or an enterprise, or you're just trusting iCloud with all of your personal photos so that you can have it scan it and try to tag things.
kevin_1_05-29-2024_101010
They need to make the case that down to the actual. [00:21:00] Chips themselves, proprietary solder. We control
kevin_1_05-29-2024_101010
everything.
gavin_1_05-29-2024_101010
they're attempting to do, I think, which they're really trying to do. Like that is what, and this is going to be, what's so interesting about WWDC is if they come out and announce that, that is a giant goal to kind of try to hit. But it does seem hard for me to believe that they could really scale that big, that fast, but more power to them if they can.
kevin_1_05-29-2024_101010
, they've been making these chips with the promise of their superpower processors for AI. So it would make sense. They're going to put a bunch of them in a server somewhere and use it to crunch data. It also helps them wean off the narrative that Nvidia is the King maker here,
kevin_1_05-29-2024_101010
right? Google has announced they have their own chips.
kevin_1_05-29-2024_101010
Microsoft's been working on their own. We know OpenAI and Sam Altman want their own, but Apple has their own
kevin_1_05-29-2024_101010
already. And so if they can show that those chips, are already crushing it, then that's a great narrative for them to have.
kevin_1_05-29-2024_101010
They don't need NVIDIA to win this AI race. So We think it's interesting. You're right. We're going to be tracking this for the next two weeks and then [00:22:00] beyond because WWDC is happening soon. And I expect , more and more leaks to happen up until then.
kevin_1_05-29-2024_101010
And then to round out the news Gavin, before we move on to the rest of our program, I want to shout out a TED talk demo that kind of set the AI world on fire last week.
kevin_1_05-29-2024_101010
GitHub is owned by Microsoft. It is a massive site where programmers can share their code and collaborate. And you can find a lot of cutting edge AI stuff there. But Thomas Domke, the CEO of the company gave a demo of co pilot workspace and you should go watch the full demo Gavin, but maybe we'll put a little clip here because what's really, really impactful is that the future development environment, according to Thomas. Looks a lot like ChatGPT right now.
23:00]
kevin_1_05-29-2024_101010
Natural language is the next programming language. You just say what you want into the box, and the box is smart enough to build the thing.
gavin_1_05-29-2024_101010
That is a very cool thing. Go check it out. That's a fun Ted talk. , and there are other things that we have seen amongst the internet that are very exciting that we want to talk quickly about. It is time for AI. See what you did there.
gavin_1_05-29-2024_101010
I saw a video, Kevin, that I was really blown away by in that it's not like the most dynamic use of AI, but what I love about this is it visualizes AI for normal people to understand.
gavin_1_05-29-2024_101010
And I really encourage everybody to go and watch this. This is a video called the Reverse Turing Experiment by Tore Knabe. I hope I have [00:24:00] your name right there, Tore. what he did was basically he created some sort of video game looking engine where it's him and first person sitting in a train car and there are four people around him and a conductor comes into the room and the game he set up basically was he was himself talking in, in voice and each of the other characters was a famous person in the visual, but actually was being represented by one of the unique LLMs in the world.
gavin_1_05-29-2024_101010
So you had GPT four. You had cloud three and a bunch of them. And the goal was each of them had to say who they were and kind of give an understanding of what they thought of AI. And then they all had to guess as to which one was not the AI.
Speaker 3
However, according to our train's Wi Fi usage, there are only four A.
Speaker 3
I. s in this compartment. One of you is a human. Huh? The human has to buy an extra ticket. A human among us? Who is it?
Speaker 8
My question. Aristotle, what if there were A. [00:25:00] I. s at the time when you came up with all the stuff that you came up with?
Speaker 5
Intriguing query, Genghis. Had artificial intelligences existed in my era, it would have profoundly expanded my contemplations on the nature of being and intelligence.
Speaker 7
I believe Genghis is the human. His response, though bold, lacked the nuance and strategic thinking I'd expect from an A.
Speaker 7
I. modeled on his conquests.
gavin_1_05-29-2024_101010
And they, you know, three of the five of them guessed he was not the AI. Which, granted, his answer was a little bit, like, choppy when he answered the with the thing.
kevin_1_05-29-2024_101010
I think his answer was almost in jest. They were all playing historical
kevin_1_05-29-2024_101010
figures, right? So you had, like, Cleopatra and Gandhi, and in this case, Tore , was personifying, ,
kevin_1_05-29-2024_101010
Genghis Khan. I
gavin_1_05-29-2024_101010
Genghis Khan.
kevin_1_05-29-2024_101010
And so his answer was just about, I think he gave a quote from Conan the Barbarian.
gavin_1_05-29-2024_101010
exactly.
kevin_1_05-29-2024_101010
was his quote.
kevin_1_05-29-2024_101010
So it was pretty obvious that he was having fun with his demo, but, Wasn't there something really bizarre about watching the one human in the room interacting with
kevin_1_05-29-2024_101010
these AIs in that way? It made me feel like this is what [00:26:00] future CAPTCHAs are gonna be.
gavin_1_05-29-2024_101010
Oh, for sure.
kevin_1_05-29-2024_101010
of robot bouncers and you wanna get in this website?
kevin_1_05-29-2024_101010
Well, prove you're over 18 and prove that you are a human.
gavin_1_05-29-2024_101010
Yes. And I think it just showed me like, okay, this is what the future of interacting with not just one AI, but multiple AIs could look like. And you know, there's all sorts of interesting sci fi ways to look at how we will interact with the eyes in general. If you haven't read the books, altered carbon, I really recommend them.
gavin_1_05-29-2024_101010
There's one of my favorites. Favorite sci fi books, but they do a really good job of personifying like AI is one AI that runs this hotel that the guy stays at. I just think this was a really kind of entry stage visualization of like what it's like to interact with multiple forms of AI. So go check it out.
kevin_1_05-29-2024_101010
It's a great video. Everybody should watch it. And if you think you've seen Altered Carbon because you watched that Netflix series, go read the books. because you have not experienced
gavin_1_05-29-2024_101010
Have you read the books? Did you read the
kevin_1_05-29-2024_101010
at all. Yeah, yeah,
gavin_1_05-29-2024_101010
The books are so much better. The books are so
kevin_1_05-29-2024_101010
So good. So now that was, , AI in a game world.
kevin_1_05-29-2024_101010
Let's talk about actually building that game world with AI Gavin. This was an AI assisted [00:27:00] game design demo that is very early stages, but still so exciting. And Evan Jones posted this on X. , the quote is, let's dream up a gothic tomb what you see on the screen, , is Evan drawing with a little crosshair, selecting portions of a gray grid. Imagine the holodeck, if you will, and saying, I want a gothic tomb. And then drawing on the wall and saying, Make this, , an archway.
kevin_1_05-29-2024_101010
And then drawing another thing and saying, Make this a gothic stone hallway. And then adding models into it. And what you really quickly realize is that this is 100 percent going to be a gothic tomb. The future of game design, , whether you're just roughing out a level or maybe even adding finishing touches, once this thing gets more feature rich,
kevin_1_05-29-2024_101010
we are all going to be game designers just by speaking things into existence.
gavin_1_05-29-2024_101010
It made me really think God, this is the killer app for the next, what may be the third iteration of Apple vision pro or whatever we're going to get,
kevin_1_05-29-2024_101010
Oh, sure.
gavin_1_05-29-2024_101010
imagine a world
kevin_1_05-29-2024_101010
Roblox and Fortnite where, you and I get to build, Collaboratively.
gavin_1_05-29-2024_101010
And that, like I would love, I could spend hours and maybe this is not a good thing because this is also where humans are going to go is like getting lost in other people's designs like this, right?
gavin_1_05-29-2024_101010
You can imagine a world where this is not good for media either, because you could imagine a world where instead of watching a show at night. You like, I'm going to go to my buddy's cemetery and go visit his experience to see what it was like. And it doesn't even have to be a full blown game.
gavin_1_05-29-2024_101010
It could be like a half an hour walkthrough of the thing that he made. Now we're talking about media, not just being, it is interactive, but it's also like exploratory.
kevin_1_05-29-2024_101010
I gotta tell you though, I did the Netflix Stranger Things Roblox integration.
kevin_1_05-29-2024_101010
Because I had a pal that, was involved with it so I went to check it out. , for a Roblox expression of a Netflix series, right? Designed for, not to grab me, I wasn't the target audience for
kevin_1_05-29-2024_101010
it, but there was a massive screen in like this shared park environment where you were watching the first episode of a season.
kevin_1_05-29-2024_101010
And then, throughout points in the episode, you could [00:29:00] Throw waffles at the screen
kevin_1_05-29-2024_101010
or a monster would attack. So it was a sort of this like immersive thing. And then in between showings, you could go around and explore the virtual environment there and get collectibles and unlock cosmetics and totally palatable, serviceable integration for stranger things within the Roblox universe. When you say about like, oh, it could be scary for the future of media, you're right, but also could be a massive boon for the
kevin_1_05-29-2024_101010
future of media when fans can go and express an IP in an environment like this. You can now, , touch users that might not
kevin_1_05-29-2024_101010
be experiencing your thing on a core platform. Maybe I don't subscribe to Peacock, but if I could go to the Trader's Mansion and play the
kevin_1_05-29-2024_101010
interactive game of it. In that environment, I would go and do that. I just get excited for the future where we can all to some level, be musicians,
kevin_1_05-29-2024_101010
be game designers, be programmers, be artists,
gavin_1_05-29-2024_101010
so I want to talk to you, Kevin, about something I did with AI this week, which was super fun. , our friends at Suno who we've had on the show before have released their 3.
gavin_1_05-29-2024_101010
5 version. So this is a new [00:30:00] version of Suno, the AI music generation app that allows people to now make songs up to four minutes in length with one shot, meaning that you can put in one prompt or one series of lyrics and make a song up to four minutes before then you had smaller chunks. You had to kind of piece together.
gavin_1_05-29-2024_101010
It's also a bridge to what they've announced, which is their 4. 0 model, part of this is probably a response to UDO's update. UDO is the kind of alternative song AI.
gavin_1_05-29-2024_101010
And honestly, in my original things, I felt UDEO had actually a better song generation engine when I played with it. UDEO also has released in painting for paid users, meaning that you can change just a small section of a song, rather than having to redo the whole song. So this feels like Suno's like, kind of like, Step to their next model, trying to roll stuff out pretty quickly.
gavin_1_05-29-2024_101010
Um, my direct experience with it was okay. I will say this. I've spent enough time with both of these, uh, AI music generation softwares now to know they are very advanced. They do a lot of really cool things. I made a song [00:31:00] for safety, Steve. I spent some, I spent some time writing these lyrics. I put it in and I generated. Quite a few of these and this is the best version that came out
kevin_1_05-29-2024_101010
AI safety, AI safety, it's time for fun with Steve. AI safety, AI safety, come along and you will see. Hey there kids it's safety, Steve. With AI rules you won't believe. We keep our AI totally in line with super duper fun, it's fine. AI safety, AI Just so I know Gavin, is he about to turn on the kids and tell them don't you ever fucking question
gavin_1_05-29-2024_101010
No, but just wait wait till there's a there's a
kevin_1_05-29-2024_101010
Okay. I just wanna make sure if I was gonna have to bleep something. Okay, I Just wanna warn the sensitive listeners safety. It's time for fun with Steve AI safety, AI safety. Come along and you'll see Steve's our super cheery, but is he qualified? We trust him with our AI care. Though his backgrounds a bit [00:32:00] unclear. AI safety, AI safety, It's time for fun with Steve, AI safety, AI safety. Hopefully no one dies.
kevin_1_05-29-2024_101010
Oh, and that's all you could ask for.
gavin_1_05-29-2024_101010
dies exactly. Okay, so
kevin_1_05-29-2024_101010
Hopefully no one dies Now, now that was one of many generations. As you mentioned,
kevin_1_05-29-2024_101010
Gavin, that was a minute long.
gavin_1_05-29-2024_101010
Now, in the past, you wouldn't have been able to generate something that long,
gavin_1_05-29-2024_101010
I did do a four minute version. I'll get to one other thing I did in a second to talk about it. I signed up for pro for this. So it is a paid account that I'm using right now. I think this is going to roll out to free users eventually, but 3.
gavin_1_05-29-2024_101010
5 is only available to pro and I think it's like 10 bucks a month for that version. And the next version, there's a higher version premieres like 25 a month. I will say if you're used to doing these on the free account, it is so much faster. Like you can get generations out now on 3. 5. It took like two to three seconds to make these generations, which is really remarkable considering it's making a whole song, right?
gavin_1_05-29-2024_101010
The thing, Kevin, that drove me a little crazy about this is in every generation I [00:33:00] tried, and I was just going to tell you all what the prompts were for this. I called it safety. Steve. I put the lyrics in myself. I wrote these in with the help of a, I think with chat, GPT, children's music, male voice, kids, TV theme song, bouncy marimba, fun, childish, right?
gavin_1_05-29-2024_101010
So those are the prompts that I put in. No matter what I did, I couldn't get that voice to, to be clear. You know, you hear that voice and it's almost like a little bit like, um, it almost feels a little bit robotic that, that kind of the voice you're hearing in the song. And I really had, I struggled as much as I possibly could to get a non robotic voice out of this.
gavin_1_05-29-2024_101010
What it seems like to me, Suno works better with is like kind of traditional song genres. Like it doesn't do the stuff that UDO does a little bit better, which is kind of like weirder jaunts to different sorts of things. I will say I tried to do this with UDO and it completely failed in a very funny way.
34:00]
gavin_1_05-29-2024_101010
So anyway, if you have Suno pro go try it, I think it's really interesting. I also believe 4. 0 is going to be a significant step up and this feels like a bridge model, more than anything else.
kevin_1_05-29-2024_101010
All right, Kevin, back in March, I posted for help redesigning our newsletter. Phil Hediatnia from Airfoil studio reached out to me and said, I got you fam. And that turned into a
kevin_1_05-29-2024_101010
massive overhaul of everything. Yeah,
kevin_1_05-29-2024_101010
So you're going to hear our chat with him about his studio, the way they integrate and don't integrate artificial intelligence into their workflow.
kevin_1_05-29-2024_101010
And I just want to caveat one thing. Phil did not ask for this.
kevin_1_05-29-2024_101010
This was not a prerequisite. He came in out of the goodness of his heart. He was a fan of the show. He said, let me help you guys out. He did not ask for a thing, which is, terrible business acumen, but very, very kind. And so I was like, Hey, listen, man, you did us a huge solid.
kevin_1_05-29-2024_101010
Please come on, talk about the process. , we want to celebrate the rebrand and we want to thank you as [00:35:00] well. So without further ado, here's the interview with Phil from airfoil studio.
gavin_2_05-21-2024_121725
We're going to do our, our intro interview question, which we always do, \ it's disturbing somewhat, but I'll ask you about how, how, how
kevin---_2_05-21-2024_121725
Your disclaimer is now far worse than the question.
gavin_2_05-21-2024_121725
Phil, we are here and it is time to answer the most important question of our time. What percentage chance do you think that AI has to kill us all?
gavin_2_05-21-2024_121725
This is a number from one to 100. And we need to be specific.
phil-hedayatnia--he-him-_1_05-21-2024_121725
Uh, 20.
gavin_2_05-21-2024_121725
Okay, that's actually
kevin---_2_05-21-2024_121725
specific. Now
gavin_2_05-21-2024_121725
specific. That's
kevin---_2_05-21-2024_121725
But it's not zero and it's not a hundred, so Now we have to justify that
phil-hedayatnia--he-him-_1_05-21-2024_121725
0 and it's not 100 because I feel like, In general, we have enough sense to know that AI should not be left to, drive things unassisted. And I think there's a lot of active conversation about it. , it's the first thing that, that governments are trying to, like, protect against.
phil-hedayatnia--he-him-_1_05-21-2024_121725
It's like, you know, letting, like, um, uh, military systems, like, just, like, make decisions on the fly over, you know, who to target, um, without, like, a human in the mix. , I thought it was interesting what, like, [00:36:00] Palmer Luckey said recently. He did an interview with Emily Chang from Bloomberg and effectively said this, right?
phil-hedayatnia--he-him-_1_05-21-2024_121725
Like, But, you know, tools are tools. There always has to be human accountability in the mix because, , if somebody uses a, a knife or a gun or an AI assisted weapon, regardless, as long as there's a person at the end of the day who's held accountable for making sure this stuff works as it ought to, right, then you, you have systems to prevent, really bad outcomes.
phil-hedayatnia--he-him-_1_05-21-2024_121725
But then there's also just like the risk that first of all, that a state gets the power to do really heinous things with AI, but then goes horribly awry in terms of, the folks running that government.
phil-hedayatnia--he-him-_1_05-21-2024_121725
I think we've seen that like that can happen.
phil-hedayatnia--he-him-_1_05-21-2024_121725
Not too long ago, I was already seeing consulting firms offer businesses, like, oh yeah, you can automate away this process entirely with AI. Even now we're not quite there, but two years ago we definitely weren't. So that's why I think there's always a chance that things can go horribly wrong.
phil-hedayatnia--he-him-_1_05-21-2024_121725
But that's also the case with almost every technological advancement that we could [00:37:00] name.
gavin_2_05-21-2024_121725
that's what I always say when you see those automatic pizza makers or automatic pancake makers It's like, you know, you think they're gonna make pancakes that you want and then you see how they break down It is a disaster. It
kevin---_2_05-21-2024_121725
It's always the It's the world's first automated burger joint and then you watch the robot arm gripping an infant's head
kevin---_2_05-21-2024_121725
listen, it might be a boring, terrible human prison of a dystopian future, yes, but at least it's gonna look good,
gavin_2_05-21-2024_121725
that's right. Exactly.
kevin---_2_05-21-2024_121725
thank you for that, by the way, because we'll probably have mentioned it in the intro, but Gavin and I were strolling down the street talking about our future and we felt some ash.
kevin---_2_05-21-2024_121725
Drop on our forehead and we looked up and there at the tip top of the ivory tower was you with a long filtered cigarette Saying
gavin_2_05-21-2024_121725
Oh, I think it was gonna be sprinkling it on us.
kevin---_2_05-21-2024_121725
Oh no No,
gavin_2_05-21-2024_121725
of your, this is the bodies of the former person. No, that's not what you're saying?
kevin---_2_05-21-2024_121725
was in his ivory design tower at the Airfoil headquarters and he looked down and he saw two enterprising young rapscallions and he said, I can help these two little kids out.
kevin---_2_05-21-2024_121725
Come on [00:38:00] up, boys. Come on up. And you graciously redesigned the AI for Humans empire, , for us. And I use the empire term incredibly loosely, as you know. Gavin and I put our thoughts together and you graciously had your team drum up some designs and rightfully you then thwacked us across the forehead and the wrist with a ruler and said, now, now children, you know, not what you speak of. Let me do something that I think evokes your brand.
kevin---_2_05-21-2024_121725
And what I got back and what Gavin got back was something that was way more playful, way more fun. And we had instantly went like, Oh yeah, yeah, that's it. So I'd love to know like what. What is the process like on your end that led you to get to that? Tell us about our own redesign because we are our favorite subject on the show.
phil-hedayatnia--he-him-_1_05-21-2024_121725
I think the biggest thing that stood out to me kind of early in the process was that you know, if the title of the show is AI for Humans and it's very much like thematically something that's accessible for everybody , it's not kind of like locked in like the ivory tower. It's not like just for people who are like building AI products, but [00:39:00] like for all of us and understand like what this tech is going to mean, then it has to be something everybody can recognize and just get.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And so, we started with a little bit more of the kind of like highbrow kind of NPR a theme, but then we wanted to throw in these like playful elements. And that's why we went with like the pixel art approach, and also just bright colors as well. I did a lot of stuff as well in like the crypto space and like crypto seems to be dark and brooding and imposing and like everything.
phil-hedayatnia--he-him-_1_05-21-2024_121725
It's like dark purple and black everywhere. It's like, no one wants that. , and so especially with AI products, I actually often encourage people to think about like light mode. I like to say the design is always an active communication. And the reason is because communication isn't just about what you want to say.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And actually I'm, by the way, I'm totally cribbing that line from a guy named Don Norman. A very, very famous designer wrote a book called The Design of Everyday Things, where he said design is an act of communication. But what that means, fundamentally, is when you communicate with somebody, you don't just think about, What do I want to say?
phil-hedayatnia--he-him-_1_05-21-2024_121725
You think about what do they need to hear? What are they going to relate to, right? What is their previous life [00:40:00] experience? Where when they hear me use a certain word or phrase, or mention something, they're going to say, Oh, I get that, I remember that. So if people, when they see the brand for like AI for humans, if they think back to like kind of , the optimism of the eighties and nineties a little bit, that's a lot better for people to think about then, you know, thinking about kind of like a dystopian future that could come to even your first question from like these advancements.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And the more it harkens back to an era that people look back on fondly, you know, the more, uh, excited people get about the tech as
kevin---_2_05-21-2024_121725
Thank you. I've been screaming this to Gavin. He's just like,
gavin_2_05-21-2024_121725
I just want killer robots. I just want grays. Yeah, exactly.
gavin_2_05-21-2024_121725
I will ask a question because the one thing I find really I love about our design is the font choice. And this is going to sound super nerdy, but it reminds me of two things kind of combined together. One is In Carta, which is a classic Microsoft encyclopedia that we used to get on CD ROMs and the Idiot's Guidebooks, which I think is a really interesting combination of our two worlds because truly we [00:41:00] are two idiots trying to understand this world.
gavin_2_05-21-2024_121725
But also in Carta is like this was this kind of thing that's like a throwback and it was like the early days of that 90s internet and you kind of felt like the possibilities were endless.
phil-hedayatnia--he-him-_1_05-21-2024_121725
I think people have picked up on that, that like, hearkening back to those products is a really effective way to convey that like, hey, this is just a different technology, but it's a similar jump, right? So I started using a computer probably way too young but what that also meant was that the first computer I used ran Windows 95 which was a version of 95 that Microsoft pre bundled with some computers to both show what the Internet could do, which I thought was really interesting, as well as, for example, it had Encarta bundled.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And I remember Encarta 98 because you would open it up and, you know, I mean, they couldn't really do fancy animations or anything like that, but you would see these images flash on the screen and they would, they did this audio mashup of all these world leaders talking and music from around the world.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And then it's like, boom, Encarta. And look, Encarta has like such a small sliver of all the information you can [00:42:00] get on the internet today, but it was presented to users as the world is now accessible to you. Isn't that so cool? And so that optimism from those days, when I visit my family, , back in Ohio, just where I grew up, , I insisted that my parents keep that computer.
phil-hedayatnia--he-him-_1_05-21-2024_121725
, and it is still there, and it still runs, and it's like, I have, like, booted up from time to time, because it was so cool to just see, when people could just imagine anything that they could do with a personal computer, and when there were plenty of doubters about the value of the personal computer, What they chose to imagine and how they chose to convey it.
phil-hedayatnia--he-him-_1_05-21-2024_121725
I think it says a lot.
gavin_2_05-21-2024_121725
That's really smart. Well, let's talk about AI in your workflow and as somebody who's a designer. I'm curious to know, , are you finding, both with you and your team, AI useful at this point?
gavin_2_05-21-2024_121725
Are you, did you use it on the rebrand of our show?
kevin---_2_05-21-2024_121725
How many humans did you displace in
gavin_2_05-21-2024_121725
that's, I wasn't getting at that, Kevin, I wasn't getting at that, but yes, I understand that question as well.
phil-hedayatnia--he-him-_1_05-21-2024_121725
In this rebrand, we didn't really use AI for anything, to be honest. I think it's important to kind of talk about also [00:43:00] like where AI can really provide value in this process. Cause I think people sometimes misunderstand the value that designers actually bring to the table.
phil-hedayatnia--he-him-_1_05-21-2024_121725
In SF for two weeks, and, , I was at this mixer, and, somebody was asking me very much that same question. It was like, so are you concerned that, like, AI is just going to displace designers? Like, eventually AI is going to just commoditize design, and, you know, we're going to be able to, like, push a button and get an interface.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And I said, no, I don't think that's true, because I think that, , neglects the value that designers bring to the table. It's like, oh, they, they create visuals, right? It was like, No, no, no, no. That's not the valuable part, right? The valuable part is deciding what the visuals ought to be. It's not the making of them part.
phil-hedayatnia--he-him-_1_05-21-2024_121725
That part is going to get easier and easier and easier to do. And there's already places where it's gotten easier for designers. So just as an example, one of the first plugins I ever used in Figma, , that provided value to me. Was there's a plugin that allows you to generate, , names.
phil-hedayatnia--he-him-_1_05-21-2024_121725
and profile pictures, , for elements that are on a page. So the reason that's relevant is that let's say you were [00:44:00] designing a page that had to have a bunch of profiles on it. What you would normally do is you'd go take 30 minutes and you'd just type in a bunch of fake names and then find a bunch of images on the internet and then drag and drop them in to like fill
kevin---_2_05-21-2024_121725
Laura Mipsum Jr. Laura Mipsum III. And it's just like a mind numbing, laborious task just to get placeholder text and images.
phil-hedayatnia--he-him-_1_05-21-2024_121725
Exactly, and then along comes this plugin that I can just press a button, select everything that it needs to fill out, and it'll just generate that text, right? That saves me a lot of time. , there's a lot of places that the design process can be made a lot more fast and efficient by doing this, but the hard part is when you have to do something that hasn't been done before, right?
phil-hedayatnia--he-him-_1_05-21-2024_121725
What's hard to do is come up with an interface that allows me to combine voice, audio, and text input into one, like multimodal input, right?
phil-hedayatnia--he-him-_1_05-21-2024_121725
That's no, one's done that before. So now we've got to invent stuff. And that's the place that designers like that's where they actually fit. So are we going to [00:45:00] get much faster to designing and building stuff because of AI? Of course. And my hope is that we get four or five times faster as a result of this.
phil-hedayatnia--he-him-_1_05-21-2024_121725
But, deciding what AI ought to do is the hard part. Letting AI do it or letting people do it. That's the easy part.
kevin---_2_05-21-2024_121725
People are talking about like a post interface world where the UI is going to dissolve. We're going to go from, , humans needing to understand computers to computers are just going to understand humans. So whether it's voice or visuals or text, whatever the input is, you're just going to be able to say it or speak it.
kevin---_2_05-21-2024_121725
The computer is going to understand it and do it. You will not need to look for a toolbar deep within a menu. You won't need to disclose a ribbon to align some text. You won't have to know how to manipulate windows necessarily. Do you think we're heading towards that? And, have your teams been experimenting with what AI might unlock in the future of interfaces?
phil-hedayatnia--he-him-_1_05-21-2024_121725
So there's definitely been some projects that we've taken on at air foil that I think are really trying to think through [00:46:00] these sorts of design challenges. , but I think it really just comes down to the fact that hands, that voice, the gestures, these are all us using relatively.
phil-hedayatnia--he-him-_1_05-21-2024_121725
, both precise, but imprecise tools, , to convey something. And so systems then are never going to perfectly understand us unless they're able to understand us by directly understanding our thinking, right? I'm not a researcher in like the brain machine interface field.
phil-hedayatnia--he-him-_1_05-21-2024_121725
And so I think there's like a lot of very interesting frontier research there. And it's very difficult for me to predict where that's going to go. But. If that does not turn out the way that we hope it does, then we're always going to need interfaces because interfaces take imprecise inputs and turn them into precise instructions.
phil-hedayatnia--he-him-_1_05-21-2024_121725
That interface might be, you know, a chat bot that we speak to, but it's still an interface, even if it's doesn't have like a bold italic underline button, like a typical word processor would
Phil, thank you so much for coming here. We appreciate it. Where can people find you in airfoil?
phil-hedayatnia--he-him-_1_05-21-2024_121725
so [00:47:00] our website is airfoil. studio. Um, and so if you're building something new and exciting and want a design partner to work on it with, we're always taking on new projects. We'd be really excited to chat.
gavin_2_05-21-2024_121725
We'll put that all in the show notes and thank you again for this amazing redesign. We are thrilled with it. I think, , both Kevin and I when we first saw it, we were like, well, that's it. It looks fantastic. So yeah, yeah.
kevin---_2_05-21-2024_121725
know! It's great to work with humans that are amazing at what they do. So thank you again for the, , the incredible redesign.
phil-hedayatnia--he-him-_1_05-21-2024_121725
Really appreciate it. Thanks guys.
gavin_1_05-29-2024_101010
All right, everybody. It's time for our five star reviews. Every week, , we go through our reviews on Apple podcasts. We have two new five star reviews this week, Kevin. I'm very excited about that. As per always, the first one is called uh, aI tack of the show.
gavin_1_05-29-2024_101010
We love this
kevin_1_05-29-2024_101010
I see what they did there.
gavin_1_05-29-2024_101010
what you did there. Big AOTS fan back in the day did a, where are they now? Search on Kevin only to find out he, he too had moved on from video games to AI loving the show and good to be introduced to Gavin through it. Thank you guys for all the effort.
gavin_1_05-29-2024_101010
That is from Tom dog. Oh, [00:48:00] two Oh seven.
kevin_1_05-29-2024_101010
Thank you, TomDog0207, for that five star review. Pleasure to have you. And I want to say, I don't think either of us have moved on from
kevin_1_05-29-2024_101010
games. we still love
gavin_1_05-29-2024_101010
I still play games all the time.
kevin_1_05-29-2024_101010
my time is up. Definitely more monopolized with AI, personally, than it is
kevin_1_05-29-2024_101010
with playing games. , JRO818, shout out to the 818, says, Best of the best, Gavin. Quote, I watch and listen to a lot of AI related podcasts and YouTube channels, and AI for Humans is hands down my favorite. I look forward to a new episode dropping every Thursday, and the little short form episodes that cover breaking news and tech conferences are like extra nuggets of awesomeness?
gavin_1_05-29-2024_101010
We love it. We
kevin_1_05-29-2024_101010
That was me eating the nuggets. I can't wait until episode 100, when Gavin reveals that Kevin was an AI co host all along. Keep up the
gavin_1_05-29-2024_101010
What if that, what if that was the case? What if I had, what if I had actually just cloned you and really you're living your best life on a mountain in Kathmandu somewhere, and I'm just using your whole thing to do this. [00:49:00] Yeah.
kevin_1_05-29-2024_101010
you can't capture this expressiveness
gavin_1_05-29-2024_101010
Well, everybody, thank you so much for listening. We will be back next week. Oh, we have a really cool video coming out on Monday where we're going to show everybody how we make our AI co host. So make sure to tune into our YouTube channel for that.
gavin_1_05-29-2024_101010
It's longer, but I think y'all will like it. We're going to go step by step on how to make a YouTube co host without code. It is a fun series, and I think we're going to have a good time doing it.
kevin_1_05-29-2024_101010
Remember, like, subscribe, engage. It genuinely helps us out. It's the only way we can grow this podcast. We do not advertise, so we actually rely on you solely. We are putting that burden on your shoulders. Thank you in advance. Five star reviews, comments, shares, likes. Thank ya. Goodbye?
Yeah. Yeah. Yeah. Yeah.