Good morning. Bloomberg reported yesterday that OpenAI is about to launch an AI tool that can use a computer … kinda similar to Anthropic’s already-coined “computer use” function.

So, everyone is working on the same thing, at the same time, and is meeting the same diminishing returns.

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

In today’s newsletter:

• ☀️ AI for Good: Streamlined solar panels 

• 👀 Greg Brockman returns to OpenAI

• 🇺🇸 OpenAI’s got a plan for US AI strategy

• 📊 ChatGPT’s impact on the labor market

AI for Good: Streamlined solar panels 

Solar panels are an important driver of the push toward clean energy generation. But the gap between the study of new solar materials and their commercialization is wide, due to the amount of testing necessary to understand the properties of a given material. 

What happened: Researchers at Johns Hopkins University developed a neural network, trained on thousands of data points from a single solar cell. 

• It eliminates the arduous task of solar cell fabrication, vastly reducing the barriers between solar innovation and adoption.

• "Instead of laboriously making multiple measurements on many devices to learn what you need to know about device behavior,” you can use this single machine learning algorithm to “tell you everything you'd want to know about a device and its properties from a single measurement that takes about 30 seconds,” Susanna Thon, a co-author of the study, said.

The researchers added that the algorithm has implications beyond solar panels, saying that it could be used to study a wide range of similar devices, including photodetectors and LEDs.

AI can now handle your phone calls.

The call center industry is on the verge of a major transformation, with generative AI leading the way.

Thoughtly reimagines phone calls with AI that speaks any language. With Thoughtly, you can build custom AI phone agents that keep your call center operating smoothly 24/7, delivering fluid and engaging interactions so your human agents can focus on the high-impact work that matters most.

From sales to support, Thoughtly’s advanced platform streamlines interactions across channels and integrates effortlessly with your existing tech stack.

Plus, Thoughtly offers a comprehensive suite of analytics, so you’ll always have insight into what’s working and where there’s room to improve.

Join the future of calling with Thoughtly (for free) today.

Greg Brockman returns to OpenAI

OpenAI co-founder Greg Brockman has returned to the company after taking a three-month leave of absence over the summer. 

The details: Bloomberg reported that, in an internal memo, Brockman said that he has been working with CEO Sam Altman on “creating a new role for him to focus on significant technical challenges.”

• Brockman’s return caps off a months-long exodus of key leadership personnel and safety researchers; this has included co-founders John Schulman and Ilya Sutskever, and more recently, CTO Mira Murati. 

• According to The Information, Mira Murati — who is launching a new startup — has already attracted some talent from within OpenAI’s own ranks. 

“Longest vacation of my life complete. Back to building,” Brockman said. 

Warp: Payroll and Compliance for Founders and Startups

In a landscape crowded with payroll solutions, Warp takes a radically different approach: do less, but do it perfectly.

While other providers compete in a features arms race, Warp strips away complexity to focus on what matters most—paying people accurately and managing compliance effortlessly.

Traditional payroll software often comes loaded with dozens of unused features, while frequently stumbling on the essentials. Warp's streamlined platform delivers exactly what modern businesses demand: lightning-fast payroll processing, two-minute employee onboarding, and automated multi-state tax compliance.

From same-day contractor payments to seamless FinCEN BOI filings, every core function is engineered for speed and precision. No bloated features. No unnecessary complexity.

Just a powerful, intuitive experience that lets startups handle their payroll operations in minutes and return to what they do best—building. In an industry that equates more features with more value, Warp proves that thoughtful simplicity is the ultimate sophistication.

Startups should spend more time in Notion and less time on the phone with the Pennsylvania Department of Revenue. Learn more about Warp and get a free AirPods 4.

• Supercharge your QA and Dev Team's testing productivity with Tricentis Copilots, generative AI-powered assistants.

  • Get started with Tricentis Copilots today.*

• Denmark launches landmark framework for using AI under EU rules — with Microsoft backing (CNBC).

• Elon Musk’s supercomputer freaked out rivals (The Information).

• YouTube is testing music remixes made by AI (The Verge).

• TikTok’s new trademark filings suggest it’s doubling down on its US business (Semafor).

• Apple’s next device is an AI wall tablet for home control (Bloomberg).

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

• Senior Director, Science & ML: Sirius XM, Lawrenceville, NJ

• Product Management Director: ADP, Parsipanny, NJ

• Shira Perlmutter, the director of the U.S. Copyright Office, confirmed in testimony before the Senate Judiciary Committee that that office will publish its findings regarding the legality of training generative AI models on copyrighted materials this year.

• GEMA, a copyright collection society for musicians, has filed suit against OpenAI, alleging the reproduction of protected song lyrics without proper payment, permission or licensing.

OpenAI’s got a plan for U.S. AI strategy

OpenAI has put together a blueprint for AI infrastructure in the U.S., according to a document reviewed by CNBC, which reported that the startup planned to present the document to lawmakers on Wednesday. 

The details: The document — which describes AI “as foundational a technology as electricity, and promising similarly distributed access and benefits” — proposes AI economic zones and a North American AI Alliance, and asks for nuclear power from the U.S. Navy.

• The document claims that investing in AI will result in enormous job and GDP growth, while also modernizing and supercharging our electric grid with nuclear power. 

• The “economic zones” are designed “to give states incentives to speed up permitting and approvals for AI infrastructure.”

Some context: It is important to keep in mind that OpenAI, a partially for-profit company in the process of becoming fully for-profit, stands to gain quite a lot here. Right now, with regulation lagging — and likely to continue to falter — there is no incentive in AI but the bottom line. 

The other element of this is that rushing to build out AI infrastructure poses myriad risks to the environment and its denizens, including spiking emissions, air pollution, grid destabilization and the reduction of natural spaces and finite resources. 

Also, as we explore below, OpenAI’s CEO has warned of the job loss that will eventually result from AI; clearly, this is not a winning position when it comes to getting the government to spend money on uncritically building out AI infrastructure.  

ChatGPT’s impact on the labor market 

Almost from the very beginning, ChatGPT — and the other generative chatbots that it inspired — has stoked a poignant fear of mass job loss, something that would seemingly lead to a dramatic restructuring of the labor market, global economies and modern society. 

This fear, in part, has been parroted by the same personas who have tasked themselves with developing AI; both Elon Musk and Sam Altman have claimed that a universal basic income, will, eventually, become a necessity due to the impending AI labor force. 

The clear challenge to this is that today’s AI — referring to the Large Language Models (LLMs) that back generative AI chatbots and systems — is bounded and limited in a few key ways, including reliability, bias and resource consumption. 

Despite this, demand for certain jobs is already falling. 

New research, conducted by a trio of business and economics professors — and first published by the Harvard Business Review — sought to examine the short-term impact of ChatGPT and generative AI on the labor market. 

The findings: By examining trends of job posts on an online freelancing platform, the researchers found “significant short-term job replacement” after the introduction of generative AI. 

• Through an analysis of more than 1.3 million job posts on the platform, the researchers found a 21% decrease in the weekly number of job posts for automation-prone industries, compared to more manual-intensive ones. 

• Writing jobs were the most affected, experiencing a 30% decrease, while software, app and web development jobs experienced a 20% decrease. 

Similarly, the researchers noticed a 17% decline in demand for graphic designers and 3D modeling freelancers following the introduction of image-generation models in June 2022. 

Following these trends over time, the professors found “no signs of demand rebounding, revealing a growing trend of job replacement.”

“It’s self-evident that generative AI competes with the work it’s trained on, and the people behind that work. But there is now a growing body of evidence to support this,” Ed Newton-Rex, the CEO of the company Fairly Trained, said in response. 

• Last month, Newton-Rex published the following statement: “The unlicensed use of creative works for training generative AI is a major, unjust threat to the livelihoods of the people behind those works, and must not be permitted.”

• It was signed by more than 35,000 people, including Kevin Bacon, Kate Bush and Kate McKinnon. 

At the same time, the researchers found an increase in job posts that listed “ChatGPT” as a skill requirement, suggesting that the “ability to integrate AI tools into work is becoming increasingly valued.”

The researchers noted that, “in the long run, there might still be net job growth as a result of AI, potentially attributed to productivity effects and reinstatement effects.” 

While this study focused on the freelance markets, such efforts at replacement are occurring within corporations as well. 

At the same time, however, separate studies have found that the integration of AI with the capacity to replace large swaths of employees will take far longer than people expect, owing to the incredible cost associated with generative AI models (a cost that is expected only to rise).

There are a few factors at play here. One is the idea of work being good enough; the data, even at this early stage, supports the idea that companies are anxious to thin out their payroll, whether or not the result is better. The last efficiency for businesses to conquer is employment, and generative AI is fundamentally a tool designed to help them conquer it. 

I don’t imagine this environment will last. Generative AI isn’t improving, but it is getting far more expensive. Eventually, the companies pushing it will no longer be able to subsidize the enormous costs of these models; at that point, they will have to charge a hell of a lot more for them, and then we’ll see if unreliable, biased systems, complete with security and privacy vulnerabilities, actually stack up to humans. 

The other side of this is that business is consumer-driven, however often consumers seem to forget this (remember the Nike sweatshop boycotts?). And there is a growing sect of consumers who are disinterested in interacting with or supporting generative AI — even if it works reliably.

People, as Dr. Noah Giansiracusa pointed out to me in July, “want to live in a world of real things.”

In the same way that people today pay more for organic produce or gourmet restaurants, why wouldn’t there be a market for businesses that are human-centric?

🤔 Your thought process:

Selected Image 1 (Left):

• “Image 2 piano didn't have back legs. You can't see how long the piano is but I feel like you should be able to see the back legs underneath or the back of the piano. It just didn't add up.”

Selected Image 2 (Right):

• “I really thought that the arms on the player in image 1 were wonky. Apparently, that's just the poor man's anatomy.”

💭 A poll before you go

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

We’ll see you in the next one.

Here’s your view on robo-surgeons:

44% of you would NOT let a robot operate on them without human oversight; 24% would.

15% might consider it, but only if it was cheaper than the alternative.

Yes(ish):

• “An operation by a robot yes, but decision to operate by an algorithm no.”

No:

• “I would not feel safe to trust a robo-surgeon yet, it currently would not have the capacity to make the complex judgment decisions that a human surgeon would. Add to that the discrimination and hallucinations issues, and I'd be worried about getting accidentally overdosed with the meds that they put you under with, or the robo-surgeon doing 'extra surgery' in areas it misidentified as anomaly(ies).”