The AI Daily Brief: Artificial Intelligence News and Analysis - GPT-Engineer Launches App as Text-to-Code Platforms On the Rise

Episode Date: December 20, 2023

Between GPT-Engineer App and v0 by Vercel, there are a growing number of text to code platforms that are expanding creativity and democratizing the capacity to build webapps. Also on this episode, NLW looks at Bill Gates AI thoughts for 2024. Today's Sponsors: Listen to the chart-topping podcast 'web3 with a16z crypto' wherever you get your podcasts or here: https://link.chtbl.com/xz5kFVEK?sid=AIBreakdown  Interested in the January AI Education Beta program? Learn more and sign up here - https://bit.ly/aibeta ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI Breakdown, we're looking at Bill Gates' end-of-year note and what he thinks about AI, innovation, and health care. Before that on the brief, the rise of text-to-code platforms. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.netnetwork for more information about our YouTube, our newsletter, and our Discord. One of the biggest themes of generative AI is the massive expansion of creative capacity. Another way to put it is the lowering the cost of creation. Now, we've seen this in the context of text creation with things like chat cheptie, image creation with things like mid-journey, which we'll be talking about a little bit later today.

And even more recently, we've been seeing a lot of video creation and music creation. Again, another theme we might hit a little bit later. Now, the common thread with all of these is that what they output is some type of media asset or multimedia asset in the case of video. They produce an image, a text, a video, a song. What they don't output is an interactive application, is a website. And yet, LLMs are increasingly good at writing code, and so it stands to reason that before long people would rip out that code creation capacity

that exists inside LLMs and inside Code Assistant apps and put it in a specific dedicated application designed to allow people to write code simply by inputting natural language prompts of what they want. Well, we have seen two projects that either launched or opened up this week that are doing exactly that. The first one we're talking about is GPT Engineer. Now, GPT Engineer has been a high-flying, exciting open-source project for the last few months. It was sort of in the family of auto-GPT-style apps

and hit on this trend of the next generation of generative AI apps, which have an expanded capacity to actually do things on their own. While now, the team behind GPT Engineer has released GPT Engineer app, and it is exactly this thing we've just been talking about, a tool to help people build web apps with natural language inputs. So, the announcement thread from Antonosika reads, Introducing GPT Engineer app. Since GPT Engineer became the world's most popular co-gen project,

I have been tinkering with the next step, how to make it practical, i.e. allow anyone to build and deploy web apps with plain English. Mission, reduce barriers to build. GPTEngineer. app lets anyone specify what you want, get a deployed web application, iterate in plain English. Crucially, we're including the co-gen community, as collaboration makes everyone move faster.

So far, we've had a blast using it to create custom animations, interactive landing pages, and Christmas-themed games. Obviously, many of the use cases are left to the community to discover. Now, if you go to gpteengineer.app, it gives you a preview of what you get if you actually get into the app. The headline reads, rapid prototyping with speed you haven't seen before. Build and iterate on interactive web apps using natural language.

You own the code and can let human developers take over at any point. It's already version controlled with Git. Now, I have signed up for access to this, and of course we'll bring you a preview as soon as I get access to it. But as I said, that's not the only tool in this sphere that we've seen launch or at least open up this week. V0.dev, which is a generative code tool from Versel, has been around for a little bit in beta, but just opened up broadly this week. This is something very similar, a web app UI code generator. So let's follow one of their suggested prompts and build a contact form. The prompt that comes up, a contact form with first name, last name, email, and message fields.

put the form in a card with a submit button. So I'm obviously not going to show you all of this, but the generation process takes a little less than a minute, depending on the complexity, and you get a set of different versions that you can build off from. Here we have A, B, and C versions with slight variations, and you can use, again, natural language input to do an updated version. So let's go back to version A and say,

make the box cyan. And there we have it, a cyan box. Now, if you're at all interested in seeing a larger demo-slash-tutorial of V0 by Verssel. I actually did one as part of the AI breakdown education beta. That program has been running throughout December and January signups are live this week. If you want to check it out, go to bit.ly slash AI beta, where there's more information and a link to register. Now, moving to another realm of the expanded capacity for generation, ever since OpenAI put Dolly 3 inside of chat GPT, people have been wondering if Mid Journey could keep up. Part of their

response has been to finally start to roll out a web-based creation experience, which is something we touched on last week, but they have also been hosting rating parties for their forthcoming V6. This is the second of those rating parties, with which after they get feedback, they'll do a final tuning, and then they'll actually release the next version of Mid Journey. From first glance, it certainly appears like the photos are another step up in terms of detail, realism, and stylistic quality, but of course we'll have to actually get in there and be able to try it out before we know that for sure. Still, this has to rate very highly on my list of most anticipated releases, so I hope we get it soon.

Now, one more quick mention that I'm not going to dwell on because I'm going to do a full video about this sometime in the next few days. Suno is an AI music app that is getting a ton of attention right now for its ability to generate the stems of real songs in a variety of different styles. Developer Nick Dobos writes, Suno is the first AI music app I've seen that can do anything close to metal, plus cringe lyrics, hilarious. Now, part of why this has gotten a lot of attention right now

is that Suno was just integrated into Microsoft copilot. or at least it's starting to roll out as a plug-in for copilot. Rowan Chung from the rundown says, Microsoft Copilot can now turn text prompts into full music tracks. It's completely free and insanely good. If you want to check it out now, and you don't have access to copilot, go to suno.aI. They give you 30 credits to start playing around with,

which equals four or five different prompts that each give you two options. Now, Suno does also offer a premium membership, which gives you much more access to create, which is, of course, how most of these AI companies are funding themselves outside of venture capital. Even Stability AI has started to bow to that pressure and has announced its paid membership tiers for commercial use of its models. There has been a lot of discussion recently about the financial health of Stability AI, as well

as a fairly open conversation from CEO Amad about what the price of different membership tiers. With the new launch, they officially have three tiers, a free tier for personal and research use, a $20 per month tier for, quote, creators, developers, and startups with less than one million in annual revenue, one million in institutional funding, and one million active users, and an enterprise plan. Basically, it's $20 per month for the vast majority of commercial usage. Now, I for one am super in support of this.

I think Stability AI pumps out models that are extremely valuable and become part of lots of other tools, not just the ones that they create, and I would like them to do whatever it takes to find financial stability, no pun intended, so that they can keep doing that good work. Someone who is not doing such good work with AI is Rite Aid. The FTC has charged and settled with Rite Aid around a, quote, biased face scanning system that led to employees falsely accusing people of crimes they did not commit. So basically what happened is that between 2012 and 2020,

Rite Aid was using facial recognition software to create a massive database of shoplifters. When one of those shoplifters walked into their store, employees would get a note, and of course in a lot of cases, the scans weren't that accurate. Now, as part of the settlement, the FTC has imposed a five-year ban on Rite Aid using facial surveillance, although whether facial surveillance will even be allowed in five years, I think, remains to be seen. This is already one of the hottest button topics when it comes to the EUAI

Act, and I think as the political conversation around AI heats up in the U.S. as well, it will likely be an important one here, too. Anyways, friends, that is going to do it for today's AI breakdown brief. Next up, the main AI breakdown. Hello, friends, one quick note before we get back to the rest of the episode. Registration for January's AI Education Beta is now officially open. It's open until just Friday at 1159 p.m. Eastern Time. You can find the link to learn more and register at b.it.l.ly slash AI Beta. Now, this is an experiment that I've been running all throughout December in which every day I drop a new video tutorial or a case study and usually partner it with a challenge, the idea of

which is to get you learning about all of these new different AI tools, as well as specific strategies for the most frequently used like ChatGBT or Dalee, and then gets you actually testing them out in the real world with real use cases and hopefully applying them back to your personal or professional pursuits as well. The first month has gone incredibly well. People seem to be really liking the video content as well as the incredible community that's forming. And part of that is that it's a group of really serious people. This is a paid experience. It's $20 a month.

Part of the reason for that is that I want you guys to judge this content on the basis of whether it's actually worth that much to you. And second, I wanted it to be full of really serious people who are intent on applying AI to their lives in some real and significant way. Anyways, I would love to have more AI breakdown listeners participate in January. Content will start on January 3rd after the end of the holidays. season, and again, the link to find out more and to register is bit.ly slash AI beta. That's bit.l.ly slash AI beta. And now back to the show. Welcome back to the AI breakdown. For today's

main episode, we're actually doing a little bit of a long read, but in the middle of the week. Bill Gates has released his annual end of year note. And surprise, surprise, the big topic is artificial intelligence. So what we're going to do is we're going to have AI NLW, read to you an excerpt first of a note from Gates back in March called The Age of AI has begun, where Gates basically gives his framework for why he thinks AI is so important, and then we're going to read, or rather AI NLW is going to read, the section from his end of year note, around specifically AI as it relates to health. After that, I'll be back for a little bit of wrap-up.

So again, we begin with the very beginning of the age of AI has begun from March 21, 2023. In my lifetime, I've seen two demonstrations of technology that struck me as revolutionary. The first time was in 1980, when I was introduced to a graphical user interface, the forerunner of every modern operating system, including Windows. I sat with the person who had shown me the demo, a brilliant programmer named Charles Simonyi, and we immediately started brainstorming about all the things we could do with such a user-friendly approach to computing. Charles eventually joined Microsoft, Windows became the backbone of Microsoft, and the thinking

we did after that demo helped set the company's agenda for the next 15 years. The second big surprise came just last year. I'd been meeting with the team from OpenAI since 2016 and was impressed by their steady progress. In mid-2020, I was so excited about their work that I gave them a challenge. Train an artificial intelligence to pass an advanced placement biology exam. Make it capable of answering questions that it hasn't been specifically trained for. I picked AP Bio because the test is more than a simple regurgitation of scientific facts. It asks you to think critically about biology. If you can do that, I said, then you'll have made a true breakthrough. I thought the challenge would keep them busy for two or three years. They finished it in

just a few months. In September, when I met with them again, I watched in awe as they asked GPT, their AI model 60 multiple choice questions from the AP bio exam, and it got 59 of them right. Then it wrote outstanding answers to six open-ended questions from the exam. We had an outside expert score of the test, and GPT got a five, the highest possible score, and the equivalent to getting an A or A-plus in a college-level biology course. Once it had aced the test, we asked it a non-scientific question. What do you say to a father with a sick child? It wrote a thoughtful answer that was probably better than most of us in the room would have given. The whole experience was stunning. I knew I had just seen the most important advance in technology since the graphical user interface. This inspired me to think about all the things

that AI can achieve in the next five to 10 years. The development of AI is as fundamental as the creation of the microprocessor, the personal computer, the internet, and the mobile phone. It will change the way people work, learn, travel, get health care, and communicate with each other. Entire industries will reorient around it. Businesses will distinguish themselves by how well they use it. Philanthropy is my full-time job these days, and I've been thinking a lot about how, in addition to helping people be more productive, AI can reduce some of the world's worst inequities. Globally, the worst inequity is in health. Five million children under the age of five die every year. That's down from 10 million two decades ago, but it's still a shockingly high number. Nearly all of

these children were born in poor countries and die of preventable causes like diarrhea or malaria. It's hard to imagine a better use of AIs than saving the lives of children. All right. And now with that prelude, let's move over to his end of your Gates note. The road ahead reaches a turning point in 2024. We're going to go to a section called AI is about to supercharge the innovation pipeline. My work has always been rooted in a core idea. Innovation is the key to progress. It's why I started Microsoft, and it's why Melinda and I started the Gates Foundation more than two decades ago. Innovation is the reason our lives have improved so much over the last century. From electricity and cars to medicine and planes, innovation has made the world better. Today, we are far more

productive because of the IT revolution. The most successful economies are driven by innovative industries that evolve to meet the needs of a changing world. My favorite innovation story, though, starts with one of my favorite statistics. Since 2000, the world has cut in half the number of children who die before the age of five. How did we do it? One key reason was innovation. Scientists came up. with new ways to make vaccines that were faster and cheaper but just as safe. They developed new delivery mechanisms that worked in the world's most remote places, which made it possible to reach more kids, and they created new vaccines that protect children from deadly diseases like rotavirus. In a world with limited resources, you have to find ways to maximize impact. Innovation is the

key to getting the most out of every dollar spent, and artificial intelligence is about to accelerate the rate of new discoveries at a pace we've never seen before. One of the biggest impacts so far is on creating new medicines. Drug discovery requires combing through massive amounts of data, and AI tools can speed up that process significantly. Some companies are already working on cancer drugs developed this way. But a key priority of the Gates Foundation in AI is ensuring these tools also address health issues that disproportionately affect the world's poorest, like AIDS, TB, and malaria. We're taking a hard look at the wide array of AI innovation in the pipeline right now, and working with our partners to use these technologies to improve lives in low

and middle-income countries. In the fall, I traveled to Senegal to meet with some of the incredible researchers doing this work and to celebrate the 20th anniversary of the Foundation's Grand Challenges Initiative. When we first launched Grand Challenges, the Gates Foundation's flagship innovation program, it had a single goal, identify the biggest problems in health and give grants to local researchers who might solve them. We asked innovators from developing countries how they would address health challenges in their communities, and then we gave them the support to make it happen. Many of the people I met in Senegal were taking on the first AI grand challenge. The foundation didn't have AI projects in mind when we first set that goal back

in 2003, but I'm always inspired by how brilliant scientists are able to take advantage of the latest technology to tackle big problems. Much of their work is in the earliest stages of development. There's a good chance we won't see any of them used widely in 2024 or even 2025. Some might not even pan out at all. The work that will be done over the next year is setting the stage for a massive technology boom later this decade. Still, it's impressive to see how much creativity is being brought to the table. Here's a small sample of some of the most ambitious questions currently being explored. Question. Can AI combat antibiotic resistance? Antibiotics are magical in their ability to end infection. But if you use them too often, pathogens can learn how to ignore them. This is called

antimicrobial resistance, or AMR. And it is a huge issue around the world, especially in Africa, which has the highest mortality rates from AMR. Nanakofi Kwaki from the Orham Institute in Ghana is working on an AI-powered tool that helps health workers prescribe antibiotics without contributing to AMR. The tool will comb through all the available information, including local clinical guidelines and health surveillance data about which pathogens are currently at risk of developing resistance in the area and make suggestions for the best drug, dosage, and duration. Question. Can AI bring personalized tutors to every student? The AI education tools being piloted today are mind-blowing because they are tailored to each individual learner. Some of them,

like Conmigo and Mathia are already remarkable, and they'll only get better in the years ahead. One of the things that excites me the most about this type of technology is the possibility of localizing it to every student, no matter where they live. For example, a team in Nairobi is working on Sominasi, an AI-based tutor that aligns with the curriculum in Kenya. The name means learn together in Swahili, and the tutor has been designed with the cultural context in mind, so it feels familiar to the students who use it. Question, can AI help treat high-risk pregnancies? A woman dies in childbirth every two minutes. That's a horrifying statistic, but I'm hopeful that AI can help. Last year, I wrote about how AI-powered ultrasounds could help identify pregnancy risks.

This year, I was excited to meet some of the researchers at Armin, who hoped to use artificial intelligence to improve the odds for new mothers in India. Their large language model will one day act as a co-pilot for health workers treating high-risk pregnancies. It can be used in both English and Telugu, and the coolest part is that it automatically adjust to the experience level of the person using it, whether you're a brand new nurse or a midwife with decades of experience. Question. Can AI help people assess their risk for HIV? For many people, talking to a doctor or nurse about their sexual history can be uncomfortable. But this information is super important for assessing risk for diseases like HIV and prescribing preventive treatments. A new South African chatbot

aims to make HIV risk assessment a lot easier. It acts like an unbiased and non-judgmental counselor who can provide around-the-clock advice. Sophie Pascoe and her team are developing, it specifically with marginalized and vulnerable populations in mind, populations that often face stigma and discrimination when seeking preventive care. Their findings suggest that this innovative approach may help more women understand their own risk and take action to protect themselves. Question. Could AI make medical information easier to access for every health worker? When you're treating a critical patient, you need quick access to their medical records

to know if they're allergic to a certain drug or have a history of heart problems. In places like Pakistan, where many people don't have any documented medical history, this is a huge problem. Mariam Mustafa's team is working on a voice-enabled mobile app that would make it a lot easier for maternal health workers in Pakistan to create medical records. It asks a series of prompts about a patient and uses the responses to fill out a standard medical record.

Arming health workers with more data will hopefully improve the country's pregnancy outcomes, which are among the worst in the world. There is a long road ahead for projects like these, significant hurdles remain, like how to scale up projects without sacrificing quality and how to provide adequate back-end access to ensure they remain functional over time,

but I'm optimistic that we will solve them. And I'm inspired to see so many researchers already thinking about how we deploy new technologies in low and middle-income countries. We can learn a lot from global health about how to make AI more equitable. The main lesson is that the product must be tailored to the people who will use it. The medical information app I mentioned is a great example. It's common for people in Pakistan to send voice notes to one another

instead of sending a text or email. So it makes sense to create an app that relies on voice commands rather than typing out long queries. And the project is being designed in Urdu, which means there won't be any translation issues. If we make smart investments now, AI can make the world a more equitable place. It can reduce or even eliminate the lag time between when the rich world gets an innovation and when the poor world does.

If I had to make a prediction, in high-income countries like the United States, I would guess that we are 18, 24 months away from significant levels of AI use by the general population. In African countries, I expect to see a comparable level of use in three years or so. That's still a gap, but it's much shorter than the lag times we've seen with other innovations. The core of the Gates Foundation's work has always been about reducing this gap through innovation. I feel like a kid on Christmas morning when I think about how AI can be used to get game-changing technologies out to the people who need them faster than ever before. This is something I'm going to spend a lot of time thinking about next year.

All right, so back to non-Robot NLW here. One of the things that I think is interesting about this is that 2024 is setting up to be a year of very fierce debate around what society wants out of AI. The voices on both the accelerationist side, as well as the safety side are getting louder. The policy and political conversations are getting more acute. And so I think it's interesting that Gates has chosen to focus on a specific manifestation of AI in terms of its usefulness in healthcare as the place where he focuses his optimism and excitement. One of the important things for this next year for these important discussions will be to measure risks against benefits. What Gates is articulating is, of course, a set of very high potential benefits that can't be ignored when considering how to approach policy in this technology.

For now, lots of food for thought, and luckily, a nice holiday break coming up to think about it. That's going to do it for today's AI breakdown. Until next time. Peace.