Good morning. This has been a week in which things just kept happening.

I had a full newsletter written, buttoned-up and ready to roll when OpenAI finally unveiled its entry into the AI search world: SearchGPT.

We break it — and a bunch of other stuff — down for you below.

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🌏 AI for Good: NASA’s latest visualization of carbon in the atmosphere

• 💻 OpenAI unveils its big Google challenger: SearchGPT

• 📄 Leaked document: Runway trained video generator on thousands of YouTube videos without permission

• 🏢 Generative AI is making workers less productive

AI for Good: NASA’s latest visualization of carbon in the atmosphere

This week, NASA published a new model that shows the movement of carbon dioxide through Earth’s atmosphere between January and March of 2020. 

The details: Because of the model’s high resolution, you can track the source of these carbon emissions (fires, power plants, cities, etc.) and watch as natural sinks absorb some of it, and the winds move the rest around. 

• The visualization was created by NASA’s Scientific Visualization Studio using a model called the Goddard Earth Observing System, which is a high-resolution weather model powered by supercomputers. 

• The model — which processed billions of data points — found that over China and the U.S., most emissions came from power plants, factories and cars. Over Africa and South America, meanwhile, emissions largely came from fires. 

Why it matters: “As policymakers and as scientists, we're trying to account for where carbon comes from and how that impacts the planet,” NASA climate scientist Lesley Ott said in a statement. “You see here how everything is interconnected by these different weather patterns.”

Struggling with graphic design? Lack the skills, budget, or time to hire a professional designer? Kittl AI is here to revolutionize your workflow.

Kittl is an intuitive yet powerful graphic design platform powered by the groundbreaking Kittl AI – the first generative AI tailored for graphic designers. Say goodbye to the limitations of static-generated images. With Kittl, you can remove and change backgrounds, fully customize colors, and integrate your designs seamlessly in just seconds, all within one platform.

What makes Kittl stand out?

• 10,000+ Professional Templates: Crafted by world-class designers for any use case—marketing assets, pitch decks, charts, infographics, and more.

• Advanced AI Tools: Streamline your design process by 75% with Prompt-to-Image, Prompt-to-Vector, Background Remover, Vectorizer, and Upscaler.

• Kittl AI Co-Pilot: Your personal design assistant, enhancing your creations in seconds.

Experience the future of design today. Try Kittl for free and use code “DEEP25” to get a 25% discount on all premium plans.

OpenAI unveils its big Google challenger: SearchGPT 

OpenAI on Thursday unveiled a prototype of its highly anticipated genAI search engine: SearchGPT. 

The details: The search engine, which has real-time internet access, is powered by OpenAI’s GPT-4 family of models. It will roll out slowly, only being made accessible to 10,000 test users at launch, The Verge’s Kylie Robison reported. It is a direct challenge to Google.

• OpenAI in a blog post said that SearchGPT will provide users with clear, relevant links to the information it returns. 

• The company said it developed the prototype with help from a number of partners, including Wall Street Journal publisher NewsCorp. 

“SearchGPT is designed to help users connect with publishers by prominently citing and linking to them in searches,” OpenAI said. “Responses have clear, in-line, named attribution and links so users know where information is coming from and can quickly engage with even more results in a sidebar with source links.”

Some context: This entry into AI search comes as scrutiny of AI search startup Perplexity has continued to increase; both Forbes and Conde Nast have accused the company of plagiarizing their work. 

• Powder, a wealthtech startup, raised $5 million in funding from groups including YCombinator.

• Lumi, an AI startup founded by Colin Kaepernick, raised $4 million in funding.

• Meta oversight board tells company to clean up rules on AI-generated pornography (Reuters).

• How Snapchat Struggled to Protect Teens From Drug Dealing, ‘Sextortion’ (The Information).

• A former Google engineer has built a search engine, webXray, that aims to find illicit online data collection and tracking (Wired).

• Dow falls from highs, Nasdaq loses earlier gains as investors sell some big tech names (CNBC).

• Las Vegas transit system is nation’s first to plan full deployment of AI surveillance system for weapons (CNBC).

Leaked document: Runway trained video generator on thousands of YouTube videos without permission

404 Media on Thursday reported that Runway, the multi-billion-dollar AI video generation company, trained its models at least partially on thousands of scraped YouTube videos and pirated movies. 404 obtained a massive internal spreadsheet that Runway used to compile videos for training purposes. 

• The spreadsheet — which can be viewed here — contains thousands of YouTube links. The largest individual spreadsheet in the document, titled “recommended channels,” includes thousands of links to specific channels. 

• The number of videos published by each channel was also listed here. 

Brands featured on the list include Apple TV, Disney, AMC Theaters and Netflix, in addition to hundreds of others (as well as content creators including Marques Brownlee). 

• Runway co-founder Anastasis Germanidis was asked by TechCrunch in June where the training data from Runway Gen-3 came from; he did not provide a direct answer. 

A former Runway employee told 404 that there was a “company-wide effort to compile videos into spreadsheets to serve as training.” They said that the company used open-source software, including YouTubeDL, to scrape the videos, purchasing proxies to avoid getting blocked by YouTube and Google. 

This all comes into an environment of copyright chaos in the AI sector, with lawsuits from media companies, authors, musicians and artists in motion, even as OpenAI and some of its peers continue to attempt to lock down content licensing deals. 

As Ed Newton-Rex, the CEO of Fairly Trained, said: “The first AI video lawsuit, when it happens, is going to be a big one.”

Invest smarter with Public and AI

If you're looking to leverage AI in your investment strategy, you need to check out Public.

The all-in-one investing platform allows you to build a portfolio of stocks, options, bonds, crypto and more, all while incorporating the latest AI technology — for high-powered performance analysis — to help you achieve your investment goals.

Join Public, and build your primary portfolio with AI-powered insights and analysis.

Generative AI is making workers less productive 

The big pitch behind generative AI is a simple one of productivity and efficiency in the enterprise. And while excitement over this pitch has led to the start of a wave of enterprise-level adoption, many obstacles — including reliability, trust, privacy and security — still remain. 

A recent study conducted by Upwork found that there is a stark mismatch between executives and employees when it comes to genAI. 

• 96% of business leaders expect genAI to make their companies more productive, and nearly half are mandating or encouraging the use of such tools. 

• And, even as 81% of said leaders acknowledge that they have increased demands on workers in the past year, nearly 80% of these workers say that the addition of genAI has added to their workload, rather than reduced it. 

The details: Upwork surveyed 2,500 C-suite business leaders, employees and freelancers across the U.S., U.K., Canada and Australia. The research broadly found that, even as executives remain sold on the productivity pitch of genAI, the reality faced by their workers is the opposite. 

• Only 26% of executives surveyed have AI training programs in place, and only 13% said they have a well-implemented AI strategy. 

• A majority of employees are also excited about the productivity benefits of genAI, but nearly half of them “have no idea how to achieve the productivity gains their employers expect.” Nearly three-quarters of employees reported feeling burned out and 65% said they’re struggling with increasing employer demands. 

Employees surveyed said that they’re now spending more time reviewing AI-generated content or learning how to use genAI tools; 21% said that they are “now being asked to do more work,” and 40% “feel their company is asking too much of them when it comes to AI.”

It’s not clear what systems/products Upwork is referring to when it says “AI.”

The AI industry has hyped the ability of its products to cheaply replace large swaths of labor, enabling companies to be more productive while cutting their payroll costs. 

We can argue about the morality of that (I don’t think it’s cool), but that’s the pitch. That’s what is driving enterprise adoption. 

The problem here is multi-fold. First and foremost, genAI is enormously expensive to both train and run. And right now, genAI startups are operating at multi-billion-dollar losses in order to entice customers to hop on board for cheap — I do not believe that this can continue for much longer. 

• So, while genAI might be cheap-ish for the user, it’s not for the developer. And at a certain point, the user is going to have to eat some of that cost. So part A of the pitch (cheap labor) likely won’t last. 

Part B of the pitch is predicated on the assumption that these models are safe, reliable, trustworthy and secure. And while we know that they aren’t, the results of this poll add a lot of credence to the manner in which that reality is being felt. 

If this trend continues, it seems possible that customer interest in enterprise genAI might falter, which could have ramifications across startups and semiconductors alike; what happens to the ‘picks and shovels’ companies when everyone else realizes there’s no gold? 

Now, this could be temporary. I, Sequoia and Goldman Sachs could be wildly wrong and genAI could become cheap, reliable and take off. But at the end of the day, a high-cost solution to a nonexistent problem just doesn’t seem like it has the kind of legs Silicon Valley wants it to have. 

A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Your view on AI startups trying to generate returns:

37% of you think that the most likely outcome here is the AI companies will get rid of free genAI access and hike the cost of subscriptions.

20% of you think that most companies won’t survive the bubble burst, so it doesn’t matter, anyway.

And a little less than 20% of you think advertising is the only option here.

^{*Public disclosure: All investing involves the risk of loss, including loss of principal. Brokerage services for US-listed, registered securities, options and bonds in a self-directed account are offered by Public Investing, Inc., member FINRA & SIPC. Cryptocurrency trading services are offered by Bakkt Crypto Solutions, LLC (NMLS ID 1828849), which is licensed to engage in virtual currency business activity by the NYSDFS. Cryptocurrency is highly speculative, involves a high degree of risk, and has the potential for loss of the entire amount of an investment. Cryptocurrency holdings are not protected by the FDIC or SIPC.}

^{Alpha is an experiment brought to you by Public Holdings, Inc. (“Public”). Alpha is an AI research tool powered by GPT-4, a generative large language model. Alpha is experimental technology and may give inaccurate or inappropriate responses. Output from Alpha should not be construed as investment research or recommendations, and should not serve as the basis for any investment decision. All Alpha output is provided “as is.” Public makes no representations or warranties with respect to the accuracy, completeness, quality, timeliness, or any other characteristic of such output. Your use of Alpha output is at your sole risk. Please independently evaluate and verify the accuracy of any such output for your own use case.}