The AI Daily Brief: Artificial Intelligence News and Analysis - New ChatGPT Features Leaked?

Episode Date: June 12, 2023

Today on The AI Breakdown, NLW looks at how the real estate industry is thinking about the impact of generative AI. Before that on the Brief, we cover a leak from potential new ChatGPT features, as well as updates from Microsoft Bing and Google Search Generative Experience (SGE).  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 discussing how generative AI is making its way into real estate. Before that on the brief, an alleged leak from chat GPT showing off new features, plus some announcements from Google and Microsoft. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Like, subscribe and share and go to breakdown.network for more information. Welcome back to the AI breakdown brief. All the AI headline news you need in five minutes or less.

We kick off today with a leak that was showing. shared on Reddit. And I will say before getting into this, that we should caveat that, of course, this is just a leak. There isn't a lot of sourcing. But this is a Reddit user who previously leaked the chat sharing feature, which did come to fruition. So make of that what you will. The screenshot shared is for something called workspace. It includes fields for files and for profile. Under files, it seems to suggest that uploading files will soon be available. And under profile, it says, add any information that you'd like chat GPT to remember about you and your preferences. Now, many people's first thought is that this is part of an upcoming

enterprise suite. We know that chat GPT is interested in the business use case. They've already announced that there will be a business version coming. Although all that we know about that version so far is that it will be private by default and is meant to answer concerns that many enterprises have about chat GPT training their models on their proprietary information. To the extent that workspace is a collaborative tool for teams or companies, it would make sense as an area that ChatGPT might want to expand into, just given how many people already use it for business use cases. Next up, we move to updates that are, in fact, confirmed. On June 9th, Microsoft announced that they were adding voice chat on desktop for Bing chatbot. Their announcement

post reads, we know many of you love using voice input for chat on mobile. It's now available on desktop by clicking on the microphone icon in the Bing chat box. Currently, Microsoft Bing voice chat supports five languages. Now, in a recent video, Microsoft Bing said that there are more than 100 million daily users of the chatbot. And in addition to this new voice chat feature for desktop, they've also announced enhanced visuals, citations, graphic-centric answers, more image-creating, chat history export, third-party plugins, multi-modality, and more. Citations are an interesting choice for them to move into, given that this is something that other chatbots like perplexity have seen some success in when it comes to how they're differentiating from other tools like

chat GPT and BingChat. Chart and graph generation seem designed to get ahead of chat GPT's upcoming code interpreter plugin, which many people are anxiously awaiting. And of course, as we've discussed numerous times, it's very clear where these chatbots and LLMs are heading, which is firmly in the direction of multimodality. As Vinnie Mora puts it, Bing Chat will understand images as well as text, allowing users to upload images and search the web for related content, such as step-by-step instructions for a craft project based on an uploaded image. Now, of course, a big part of the reason that Microsoft is trying to move so fast with their Bing chat experience is that Google's search generative experience,

answer SGE is coming right down the pipe as well. On June 8th, Google Search Liaison tweeted, two weeks ago we started opening up access to SGEE. Since then, we've made a number of quality updates, including a major improvement that reduces the time it takes to generate AI snapshots by half. We hope everyone testing it out enjoys the speedier experience. Now, this is a more important update than it maybe even seems at first, given that many people's reactions to this first version of the SGE was that it was too slow. Sure enough, here's the verge saying Google's AI-powered search experience is way too slow. Why wait for an AI-generated summary when I can just scroll down the page? Now, the interesting thing that the Verge is pointing out

here implicitly is that the AI-generated search experience is competing with the efficacy of the traditional search experience that we already know. And to do so, there's not just a content consideration, but also a speed consideration. Now, it also appears that Google might be getting ready to roll out the search generative experience to a larger group, given that they've been releasing content such as this YouTube short that's playing now about how people can sign up for access. If all of this screams AI arms raised to you, you're not alone. However, the CEO of Google is trying to shift that narrative. Sundar Pichai sat down with Bloomberg Business Week and tried to suggest that Google has been positioning for the AI rush for a long time and that they weren't going

to get caught up in some sort of inexorable tit for tat that leads them to do things that are unsafe. For example, when Bloomberg asked, does Google see this as an existential competition? Sundar said, it's a competitive moment, but I've built the company to be AI native for a long time. I feel better positioned for this than we were for the shift to mobile. Bloomberg also asked, you've said tech companies must take care not to see AI as just a race, but there's a very real sprint to market at Google and elsewhere. What are you doing to make sure ethics don't fall by the wayside? Sundara said, we've been cautious.

There are areas where we've chosen not to be the first to put a product out. We've set up good structures around responsible AI. You will continue to see us take our time. Of course, the context at least a little bit for some of these statements, is former Googler Jeffrey Hinton now being out on a media tour, saying that he's worried that Google, the company that he worked for for over a decade after selling his startup to them, was no longer exercising this sort of prudence

because of the growing financial pressure to compete with startups like OpenAI as well as established players like Microsoft. And to the extent that regulation is a part of the answer to whatever AI arms race does exist, different jurisdictions around the world are positioning themselves to be leaders in exactly that. British Prime Minister Rishi Sunak, who before becoming Prime Minister said that he wanted Britain to be the head of Web 3, is now pitching Britain as the perfect place to anchor global AI regulation.

At a recent London Tech Week conference, he said, I want to make the UK not just the intellectual home, but the geographical home of global AI safety regulation. At the same event, his finance minister, Jeremy Hunt said, I would like this country to be a force for good in making sure these are set in a way that means that technology can really benefit everyone. Meanwhile, in the U.S., on Thursday of last week, two new bipartisan bills were introduced that deal with different aspects of AI. Interestingly, neither of them is big comprehensive legislation. Instead, they are very discreet and specific. To be honest, that probably means that their sponsors actually want them to go through, as big banner comprehensive legislation is often just flag planting for the people who sponsor the bills, versus something that actually has a chance to make it through, at least in any sort of short order. Of the two bills introduced last week, one is focused on U.S. government transparent.

requiring the U.S. government to be clear when it's using AI to interact with people, and the other is focused on global competitiveness, establishing a U.S. office to determine if the United States is remaining competitive when it comes to AI. You'll remember last week we reported that Senate Majority Leader Chuck Schumer has scheduled three hearings upcoming for the Senate around AI, with the third being a classified hearing for a small group of senators only. So far, those hearings do not have a date. All right, friends, that is it for the AI breakdown brief.

If you're enjoying this, please like, subscribe and share. Remember, you can get it as a newsletter at the AI breakdown.bohive.com. That's B-E-E-H-I-I-V, and I will be back soon with the main AI breakdown. Today on the AI breakdown, we're talking about how AI is transforming real estate. When we talk about AI transformation, we're really talking about three separate things. The first is how AI is changing, how individuals live their lives, how they do their work, what jobs they do. On the other end of the spectrum, there is the societal. level disruptions of AI in terms of its impact on democracy or even the way that we conceptualize

our relationship to ourselves. But then in the middle, there is the impact on individual industries. While of course some will be more or less affected, almost every industry is now asking the question of how AI is going to impact what they do, and real estate is no exception. Now, what's interesting about real estate is that this is a conversation that's been happening for some time. Take, for example, this CNBC piece. It's titled, artificial intelligence is taking over real estate, and here's what that means for home buyers. The key points, they say, real estate companies are increasingly using artificial intelligence in every aspect of buying, selling, and home financing.

Algorithms can now go through millions of documents in seconds, looking through property values, debt levels, home renovations, and even some of a homeowner's personal information, said Compass's chief technology officer, the traditional agent would go knock on the doors of a lot of homes. Now AI helps you find the homes that are most likely to sell in the next 12 months. Now, what's interesting about this is that this was published on Friday, September 17, 2021, almost two years ago. And so I think it's interesting to ask, given that the real estate AI conversation has started over the last few years, how are people within the real estate industry viewing this new wave of generative AI specifically? Recently, Real Estate VC, Zach Arons of the company MetaProp, wrote a piece called the impact of generative AI in real estate.

One of the things that's notable about Zach's piece is that he is not focused on some future use cases. he's focused on the here and now. He writes, while there are numerous AI models available and new ones emerging, the most valuable for the real estate industry in the near term are text and image models. The first and biggest category of opportunity that Zach sees when it comes to real estate is around marketing applications. He writes, AI can be used to automatically generate property descriptions, virtual home tours, and even 3D floor plans. This saves real estate agents and marketers a significant amount of time and resources while also providing potential buyers

with more immersive and interactive experiences. Starting with text generation, Zach writes, The early adopters in the real estate industry are already using these tools to accomplish their work in a fraction of the time it would normally take to write content. However, it's not just the time saved, it's also the fact that new agents and property managers can easily produce high quality and more accurate content about their properties,

which can be difficult to do manually without significant effort and experience. By way of example, he shows utilizing ChatchipT to create a listing. He prompts, write a listing for a three-bedroom two-bathroom home with open floor plan, updated kitchen, and a two-car garage in a small subdivision. ChatGPT responds, welcome to your dream home. This stunning three-bedroom two-bathroom house is located in a quiet and peaceful subdivision. As you step inside, you're greeted with an open floor plan that's perfect for entertaining guests or spending quality time with family. The living room features a cozy fireplace, high ceilings, and large windows that led in plenty of natural light.

The master bedroom is a true oasis with a large walk-in closet and an in suite bathroom featuring a soaking, tub, separate shower, and dual sinks. Now, notably, he hasn't necessarily given it all that information, and so whoever was using this would have to perhaps customize or cut out parts that weren't exactly true or that were hallucinated. But you get how there is a marketing tonal ease here that's going to draw someone in in a way that a standard real estate listing might not. Now, on top of just generating listing, Zach also says that text generated AI could be used to personalize content, such as creating emails or property descriptions for potential buyers that are tailored to their preferences and search behavior. He says that things like chat chip BT could

be used to analyze property and market trends and create informative and persuasive content speaking to those trends that could help a buyer understand the market that they're buying in. And he points out these tools are really good at creating multilingual content. So if an agent is working in a diverse market or with international buyers, that could be a huge problem solver and time solver there as well. Now, images are potentially even more disruptive. The first way that Zach points to is virtual staging. In other words, showing how an empty space might look with different furnishings and decor. This he points out is much cheaper and faster than traditional staging. Of course, there are also a new suite of photo modification tools that can be

extremely useful in editing property photos and making them look more appealing. The example he used here has a photo of a house that has a very kind of dreary looking gray background, but he went to autoenhanced.aI and used their sky replacement feature to give it a bright blue, wispy cloud kind of view that looks much more inviting. Now, autoenhanced.comi, is a great example of a type of general tool that's now being retrofitted for a specific industry. From a toolset perspective, it's very similar to other photo modification or enhancement tools out there, but Autoenhanced.ai is clearly going for and thinking about this real estate use case. They advertise instant real estate photo editing powered by artificial intelligence.

They promise list properties faster, increase sale price by 2%, and grow online viewings by 61%. In addition to that sky replacement tool that I just mentioned before, they also have a perspective correction tool that automatically fixes wonky angles, a 360 enhancements tool that improves 360 degree walkthroughs, an auto-hDR merging tool that combines multiple HDR images into one, and a number of other tools as well. Now, beyond just photo editing, Zach is also pointing to generative AI being used to create realistic simulations of different materials and finishes. The example given shows what a kitchen would look like if it had marble countertops and a new color painted on its cabinets. A few other examples of how image generation tools could be used that Zach gives include virtual property tours, design visualizations and space planning, all of which can help architects and agents before properties are completed or even started, and provide, as he puts it, a more immersive and interactive experience than traditional 2D blueprints or renderings. Market.a.i is an example he points to that advertises themselves as generative design,

for residential planning. The company writes, market empowers architects, designers, builders, contractors, and developers to generate residential floor plans, navigate zoning codes, and explore limitless styles. The value proposition they say is generate concepts in minutes, not months. Stop spending months on schematic designs, input the constraints through parameters or natural language, and generate hundreds of residential floor plan variations instantly. So let's try to understand a couple of the themes that are running throughout a lot of these different ideas. One is increasing the quality of the content surrounding real estate transactions, writing better listings, having better photographs, etc. A second is about personalization and preference. For many, if not most real estate buyers,

especially when it comes to residential real estate, buying a house or an apartment or a condo is the biggest purchase they're ever going to make. These tools that give agents and sellers the ability to better help those buyers understand how their personal preferences might be borne out in that particular property. Obviously, is going to lead not only to more sales, but probably better sales with more satisfied customers who actually understand what they're getting into. Now, that gets to the second big category of opportunities that Zach sees, which he calls conversational applications. Some of the benefits he sees from these applications are still in the sales process. For example, he says that chatbots could be programmed to answer common questions from potential buyers or renters,

such as property details, pricing and availability. This, he says, could free up a real estate agent's time to focus on more complex issues. And he also thinks that chatbots could be used to qualify leads. Chatbots, however, also potentially, have implications for once a property is purchased or rented. Examples he gives include rent payment, such as setting up automatic payments or giving reminders when rent is due, maintenance requests to help people get faster and more efficient service, mortgage or insurance recommendations, and more. A third category that Zach discusses are search applications, specifically made possible by semantic search. Zach writes, traditionally property search engines rely on simple keyword matching, which can be limiting and imprecise. With semantic search, however, users can search for properties

based on a wide range of criteria, such as location, size, price, and specific features or amenities. For example, a user could search for three-bedroom apartment with a pool in downtown New York, and the search engine would return results that match those criteria, even if they don't include the exact words in the search query. Zach also points out that semantic search could be good at wading through the voluminous amount of content that real estate companies generate, and that it could also help enhance the accuracy and relevancy of property valuation. Now, Zach is far from the only person to have written a how AI will transform real estate article. Entrepreneur.com published something similar in February this year, and they point to a slightly

different set of potential benefits. The first, they call predictive maintenance. They write, by analyzing sensor data from properties, AI could estimate exactly when care might be required and even go ahead and schedule the work itself. Another, they say, is home automation. And a third is AI market analysis, basically helping real estate investors crunch even more information than they could before to better understand what areas might be valuable in the future. Now, this piece by entrepreneur also points out that there are some risks here, particularly when it comes to privacy concerns among tenants. Tenants, they write, may not be happy with AI monitoring their patterns of behavior. Now, while not everyone in real estate is rushing out to retrofit everything by AI,

there is certainly a lot of interest in the space. Just last week, Elise AI announced $35 million in new funding for what they call the all-in-one AI platform for your property. Around since 2017, a lease is an AI management tool that helps automate leasing, resident conversation, service request, payments, and more. If you do a search for AI real estate on Twitter, you'll find examples like Strip Mall Guy, who writes, we hear about the impact of artificial intelligence a lot, and the other day I started thinking about what a real estate transaction would look like if we used AI. He goes through how it might impact the purchase contract, the title report, parking analysis, dry cleaners, vacancy history, financing, loan documents, resolutions,

taxes, escrow fees, appraisal, and more. His conclusion, the implications of this are endless. We're entering an entirely new industry and it will come fast. Like I said at the top of the show, every industry right now is just racing to figure out what AI and generative AI specifically means for it. What I think is valuable about going and seeing what people in those spaces are discussing is that it helps us get away from the natural optimism and sometimes over-exuberance of the entrepreneur class and the investor class who's looking at AI as a whole. Hopefully this was interesting to you, even if you're not in the real estate space. And if you are, I'd love to know what you think will be legitimate and valuable uses of AI versus more trouble than it's actually. worth. Also, let me know if this type of industry analysis is something that you enjoy or is a little bit too far outside of what you're actually interested in. In either case, I appreciate you watching or

listening. If you're enjoying it, please like, subscribe, and share. Check out the podcast and the newsletter version. And until next time, peace.