Eyes, Ears, Hands and Mouth

Speaker 3

Baby, that was the week

Speaker 1

pretty much after I finish this. And you can see from the weird artwork that inevitably Andrew would comment on more this week than almost any other week is my guess. Eyes, ears, hands and mouth, the next stage in AI. And of course, what we're talking about here is the announcements this week that were made by both OpenAI and also by Google. And the newsletter is dominated by that, as you might imagine. as is the media, by the way, if you look at this.

Speaker 4

AI has moved into a new era, from chatbots to something out of the movie, Her. Good morning, Theodore. Morning. You have a meeting in five minutes. You want to try getting out of bed? Google and OpenAI, both debuting AI assistants that can emote. Whoops, I got too excited. Reason. That image of a skull reminds me of Hamlet. Make jokes. A lullaby about majestic potatoes. Now that's what I call a mashup.

Speaker 2

Translate.

Speaker 4

The objects you're showing are una manzana and un plátano in Spanish. Even remember where you left your glasses. Do you remember where you saw my glasses?

Speaker 4

Yes, I do. Your glasses were on the desk near a red apple.

Speaker 1

There you go. That's CNBC just walking through some of what's possible.

Speaker 1

She talked about emoting and remembering, but for me, the biggest part of the announcement was... that every phone on the planet is now a pair of eyes for ChatGPT and every microphone in every phone on the planet is now a set of ears for ChatGPT. And ChatGPT can now use those eyes and ears to observe both real-time video-based content that you point the camera at or still images or anything capable of being heard and listened to. which means that the cloud device that is open AI now has access to everything. Data is no longer web scraping. Data is anything that happens anywhere, including, by the way, pointing your camera at documents and being able to read them, newspapers and being able to read the magazines and so on, statues, famous places. It's quite capable of doing lots of things. A few examples here which are kind of really amazing. Let's start by looking at the real-time translation.

Speaker 2

We asked the live audience on X to submit a few requests for what they would like us to try out here. So I will take a couple of prompts. Okay, Bot Gascar wants to know if GPT-4-0 is capable of real-time translation. Mark, you want to try this one?

Speaker 4

Sure, yeah, let's do it.

Speaker 2

I speak Italian, so we can just do English-Italian.

Speaker 4

Okay, cool, let's try to do English-Italian. Sure, let's do it.

Speaker 4

Hey, ChatGPT, how are you? I'm doing fantastic. Thanks for asking. How about you? How's your day going? I'm doing great. So I would like you to function as a translator. I have a friend here who only speaks Italian, and I only speak English. And every time you hear English, I want you to translate it to Italian. And if you hear Italian, I want you to translate it back to English. Is that good? Perfetto.

Speaker 2

Mark, io mi chiedo se le ballene potessero parlare, cosa ci direbbero?

Speaker 4

Mike, she wonders if whales could talk, what would they tell us? They might ask, how do we solve linear equations? Certainly, yes. Great.

Speaker 1

Pretty amazing. That goes on for a little while longer. You can find all of those videos on the OpenAI YouTube channel. And so that's listening, but also being able to be intelligent about what you hear, including real-time translation into a bunch of different languages. What about seeing? Is the king in residence at the moment?

Speaker 4

Yes, it looks like the king is in residence right now. The royal standard flag is flying above Buckingham Palace, which is the signal that the monarch is present. How exciting to be there while the king is home. Try and tell me exactly what they're doing right now, please. Right now, the ducks are gently gliding across the water. They're moving in a fairly relaxed manner, not in a hurry. Occasionally, one of them will dip its head under the water. probably looking for food, and then pop back up. I even know when a taxi is coming with its orange light on. I think I'll hail it to get home. Yes, I spotted one just now. It's heading your way on the left side of the road. Get ready to wave it down.

Speaker 1

Those are just a few examples, but the ability to capture live video, understand what it is and help the user act on what it sees or understand or learn from is taking a massive step forward. And you can also hear the delay between seeing and hearing.

Speaker 1

are being asked and answering is non-existent. It's as if you've got somebody there with you, which is very different to the last version.

Speaker 1

Let's look at the ability of the new desktop app to interact with what you're doing on your desktop.

Speaker 4

So let me take out a computer for this.

Speaker 4

So I'm going to open the ChatGPT desktop app like Mira was talking about before.

Speaker 4

OK. And to give a bit of background on what's going on, so here we have a computer. And on the screen, we have some code. And then the ChatGPT voice app is on the right. So ChatGPT will be able to hear me, but it can't see anything on the screen. So I'm going to highlight the code, Command-C it, and then that will send it to ChatGPT. And then I'm going to talk about the code to ChatGPT. Hey, ChatGPT. Hey, there. How's it going? Yeah, it's going really well. I was wondering if you could help me with a coding problem today. Of course. I'd love to help you out. What's the coding problem you're dealing with? OK. I'm going to share with you some code. One second. Sure thing. Take your time.

Speaker 4

OK. So I just shared some code with you. Could you give me a really brief one-sentence description of what's going on in the code? This code fetches daily weather data for a specific location and time period, smooths the temperature data using a rolling average, annotates a significant weather event on the resulting plot, and then displays the plot with the average, minimum, and maximum temperatures over the year. Awesome. Thanks so much. And in the code, there's this function foo. Can you describe to me what the plot would look like if I had versus didn't have this function? Sure. The function foo xy takes in a data frame x containing temperature data and a window size y for calculating the rolling mean. It smooths the average temperature, minimum temperature, and maximum temperature columns by applying a rolling mean over the specified window. With the foo function, the plot would show smoother temperature lines because the rolling mean reduces the noise or fluctuations in the temperature.

Speaker 1

So that's probably enough of that, but you can see that the desktop app allows ChatGPT to see your desktop. Your desktop becomes an input and any content on your desktop can be used as an input as well. So suddenly its ability to help you in any given situation is increased massively. So those are huge steps forward in capability. And when I thought about this week's title, Eyes, Ears, Hands and Mouths, It was really to separate out these things to allow us to imagine what would happen if a physical object like a robot was given these attributes linked to ChatGPT and could use cameras and microphones to engage with and interact with the real world. And we're getting closer and closer to that. It's becoming, you know, very real prospect, especially as the mechanical side of robots is so improved now. Balance, the ability to deal with surroundings that are not predictable in various ways. So this is big. Google had its own event and it was the day after and I will say that the Google event was somewhat underwhelming compared to the to the OpenAI event just because Google's quite far behind and they are trying to move in similar directions and there was evidence that they made progress against those directions but they're seriously far behind with a couple of essays in this week's newsletter. One from O'Malick who talks about why OpenAI was better.

Speaker 1

My wife just asked me what burger I want for dinner. So there you go. I just need to tell her I'm recording.

Speaker 1

So, yeah, the Google event, Oum's got a great essay. That's really just the difference between a company that launched a product and a company that launched a bunch of technologies. And it really is true what he says. There was not no comparison, really, between the two. And, you know, it's also the case that...

Speaker 1

The chat GPT announcements have been widely responded to, but Peter Nixey makes the point in this week's newsletter that they're probably widely underestimated in terms of their likely future impact. His exact title is that OpenAI Spring Release will end up being far more significant than most of us might suspect. And that's because it really does take AI to a much more human place where more people can use it. Probably the biggest thing of all is that it's free. It used to cost $20 a month to get some of the pro features. uh it's now free so everyone with a with a smartphone in the world um will be able to use it without paying and that that is phenomenal i mean in any context you could really think of any context but education to name one children being able to learn is going to be seriously enhanced

Speaker 1

The other thing I noticed is that there is a lot of capability now to blend AIs. I've mentioned before on the show that I use Claude.ai, which is the anthropic version. And Claude.ai is super clever. It was, I think, better than OpenAI for some things, but that has now changed.

Speaker 1

Nonetheless, ChatGPT isn't good at everything still. And so this week I started a process in OpenAI, but I finished it using Claude. And what it was is I was trying to make a chart The chart is basically a chart showing the correlation between the NASDAQ, the S&P 500, and then within my work context, the single rank. algorithm series B rounds and how they perform over time and so when I use ChatGPT it didn't have the Nasdaq data I wasn't able to get the Nasdaq data so I went to Claude.ai and I got that data from Claude.ai and then went back here you can see it I went back to ChatGPT gave it the data. And between the two of them, they were able to produce this chart that shows that if you bought an S&P 500 share in 2014, five years later, you would have made 139% of your original investment. On the NASDAQ, you would have made 189%. And with single rank, you would have made 431% and so on and so forth each year. I couldn't have made that graph prior, you know, without using both AIs. So, you know, that in and of itself is significant. I don't know how to take that away now. Oh yeah, there we go. Can I do that? Let me, there we go, we can make it small. I've got some kind of a glitch going on, there we are. So that was two AIs acting as independent agents doing different parts of the work to produce a chart. It's getting to the point where that will happen more and more. There's lots of YouTube videos of people stringing together, you know, lots of agents all doing different parts of a task together. And that is probably the next phase. As things get better, they'll also connect with each other to produce things for human beings. The rest of this week's newsletter covers lots of things. One of them, a big deal. There's a company we've talked about called Destiny. Its ticket symbol is DXYZ.

Speaker 1

has some things in common with SignalRank in that it is a company that owns stocks in private companies. In the case of Destiny, it's SpaceX and several other very well-known stocks that it bought quite late in the life of the company. for very high prices, whereas SignalRank buys high scoring Series B investments earlier in the life for the right price, the price at the time. But apart from that, they're quite similar. It's a publicly listed vehicle whose asset is private companies. And it went public a few weeks ago. The value of the assets it owns are about $50 million in round numbers. it traded initially at a billion dollars today it trades at about 160 million dollars so you know three to four times the value of the assets that it's holding and it does so by buying shares from employees of private companies and Those private companies have now objected to the fact that it bought those shares on what's known as a forward contract. Read the article in the essays of the week, but it's making the point that there is a lot of tension since Silicon Valley over employees selling stocks to companies like Destiny without permission of the company whose stock it is.

Speaker 1

And that is going to play out over the next period. Not bad news for Zingarec because we do not buy stocks from employees and we don't buy common shares at all. We buy preferred shares in proper funding rounds. So a little bit different for us, but that is there. There's also a great article by Andre Retterath. He runs Data Driven VC and he's an investor at Early Bird VC in Germany. I spoke at his conference a couple of weeks ago. He's produced an analysis this week of all the databases that follow startups. Very good analysis. digs into Crunchbase, CB Insights, Dealroom, PitchBook, Prequin, and all of the others. And it says what they're good at and what they're less good at. Very useful for any of you who are investors. In terms of startup of the week, it has to be ChatGPT, but I focused on a specific thing they launched. If you have a team account, ChatGPT, you can now link to your Google Drive or your Microsoft OneDrive, and it can talk directly to documents, spreadsheets, slide presentations, without you having to upload them into ChatGPT. I think that's a huge advantage. change in the ability for you to do analytics and suggest improvements to real documents. And the tweet of the week is Sam Altman's tweet, acknowledging that Ilya, his co-founder, one of the co-founders, will be leaving OpenAI. Of course, there's a big controversy a few weeks ago about Sam being the CEO. Ilya was in the middle of that. Nobody really knew what was going to happen. Well, it's happened. They've now parted ways and Ilya will be moving on to a new project. That said, that is the end. That was the week for May the 17th. The newsletter will be going out shortly and Andrew will be back next week. Thank you so much for listening and watching.