Good morning. Hope you all enjoyed the long weekend.

Now that we’re post-Labor Day, welcome to Autumn; as Albert Camus put it, that “second spring when every leaf is a flower.”

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

In today’s newsletter:

• 🦷 AI for Good: A tool for periodontists

• 🚘 China’s self-driving push

• 🖥️ The UK’s first AI-driven, teacherless classroom

• 💼 Labor x AI: Union wins, automation & corporate incentives

AI for Good: A tool for periodontists

You’ve got your traditional dentists (the ones you're supposed to see twice a year), your orthodontists (the ones who gave you braces when you were 12) and then your periodontists, the specialists you bring in when your dentist starts noticing issues of bone and gum loss (also known as periodontitis). 

Periodontitis — among the most common diseases in the world — is difficult for clinicians to accurately recognize and diagnose for a number of reasons, partially because of the complex array of factors that can contribute to it. 

The details: Over the past several years, research into deep learning (specifically at the intersection of periodontitis) has proven promising. 

• Researchers have found that, in processing and analyzing patient radiographs, deep learning algorithms can identify changes in bone tissue, predicting which teeth might eventually become compromised. 

Why it matters: Combined with clinical evaluations, such an approach can lead to early detection, which, in turn, allows for early intervention. As with many illnesses, early detection is a critical first step to prevent further damage. 

Deploy AI securely with Vanta

If you’re building or selling AI-powered products, demonstrating top-notch security practices and establishing trust is more important than ever.

• With Vanta, you can quickly and easily demonstrate compliance with gold-standard AI frameworks like ISO 42001 and NIST AI RMF.

• Vanta helps you prove secure deployment of AI, build customer trust, and accelerate your sales cycle.

Plus, with Vanta’s Questionnaire Automation, you can automate your responses to lengthy security questionnaires about your security posture and AI practices.

Learn more and claim a special offer of $1,000 off Vanta at the link below.

Get $1,000 off Vanta

China’s self-driving push

Last week, China’s Ministry of Public Security revealed that it has granted a total of 16,000 test licenses for autonomous vehicles; the country has also opened 32,000 kilometers of roads across the country for autonomous vehicle testing. 

The landscape: In June, China approved nine domestic automakers — including BYD — to test their autonomous tech on public roads. 

• China is testing more autonomous vehicles than any other country; in the U.S., robotaxi experiments are only being run by a few companies in a handful of cities. 

• And while researcher concern over the legitimate viability of self-driving cars persists, in China, human drivers are worried about losing their livelihoods to robotaxis. 

Counterpoint Research’s Mohit Sharma said in a recent report that these driving jobs won’t disappear overnight. “It will be a slow transition phase area by area, region by region,” he said, adding that there will be new jobs on the other side, such as data annotation and remote vehicle monitoring. 

AI Video Assistant: Make Pro Videos in Seconds (Free Trial)

Tired of spending hours on video editing? PlayPlay's AI Video Assistant streamlines your workflow, letting you create professional videos in seconds.

• Save time: No more editing hassles.

• Create impactful videos: Our AI generates engaging visuals and compelling content.

• Streamline workflows: Easily create consistent, branded videos.

• Boost engagement: Captivate your audience with high-quality video content.

Trusted by 3,000+ brands like Dell, CVS Health, and L'Oréal.

Start creating engaging videos now – start your 7-day free trial!

• Want to become an AI consultant? Deep View readers get early access to The AI Consultancy Project. Request early access.*

• Enterprise call center startup Bland AI raised $16 million in Series A funding.

• The founders of industrial robotics company Covariant and about 25% of its staff are joining Amazon (The Information).

• Twitter blocked in Brazil after supreme court panel upholds suspension order (CNBC).

• Untold History of AI: Invisible Women Programmed America's First Electronic Computer (IEEE Spectrum).

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.

The UK’s first AI-driven, teacherless classroom 

Beginning this month, David Game College, a private college in London, is launching an “alternate form of study” with “all core subjects being taught entirely by AI-driven adaptive learning platforms.” 20 students, between the ages of 15 and 17, will partake in the pilot, which costs around £27,000 per year.

The details: The intention is to enable students to progress at their own pace, with each receiving individual support from “learning coaches” as needed. 

• The core idea on display here is “adaptive learning,” where the algorithm behind the platform regularly assesses each student’s grasp of the curriculum, then creates targeted pathways to help that student gain the knowledge or expertise needed. 

• David Game said the program is well-suited for “non-traditional learners.”

Some context: The idea of AI-powered adaptive learning has been employed successfully in some targeted channels, for instance, in the form of a generative game — called Dysolve — that helps kids with dyslexia. 

My view: When targeted, adaptive learning can be great. But outsourcing all education to generative AI, when we know such systems have issues with reliability, hallucination, and, more generally, ethics, seems shortsighted. As Chris McGovern, who runs the Campaign for Real Education, told Sky News, this dehumanizes the process of learning.

Union wins, automation and corporate incentives

This Labor Day, I figured we’d take a look at that fraught intersection of generative AI and human labor.

One of the most persistent AI-related fears has to do with job loss. Plenty of reports have surfaced in the past year that seem to add weight to this fear. A combination of investment banks, consulting agencies and corporations themselves have suggested that worker replacement is not only the future, but has already begun. 

And we are seeing this in many cases; jobs for copywriters and illustrators alike have been drying up for the past 18 months as increasing numbers of companies turn to generative chatbots as a cheaper, ‘good enough’ alternative.

But in some ways, things are not as bleak as it seems. 

• For one thing, some of the places where this integration is happening are historically understaffed. 

• In computer and data science, for instance, GenAI is helping engineers, not replacing them. Part of this is due to the fact that there has been an ongoing shortage of software engineers for years. The other part is due to the sheer demand for software engineering work. 

Plus, a June study by the Dallas Fed found that, among firms that have adopted AI, there has been no noticeable impact on employment (so far). An MIT study further found that the adoption of the kind of AI needed to replace workers, if it comes at all, will take a very long time.

• And, according to the monthly Challenger Report on job cuts, companies haven’t cited AI as the dark specter behind job cuts since April, when 800 jobs were lost due to AI (that number marked the highest monthly total since the 3,900 cuts Challenger recorded in May 2023).

That’s not to say everything is fine. The current tech is well-designed to ‘empower’ employers to reduce their headcount. Part of the reason behind this has to do with corporate incentives. 

• A 2020 paper written by a team of economists — including MIT economics professor Daren Acemoglu — found that the U.S. tax code “systematically favors investments in robots and software over investments in people.”

• As of 2020, the effective labor tax rate stood above 28%, while the tax rate of capital invested in equipment and software declined to around 5%, according to the paper. 

The answer isn’t as easy as swapping the two numbers, according to the researchers, since automation isn’t always a bad thing. The idea is just to shift the incentives so “that you are leaving automation decisions in the private sector but you are getting rid of artificial incentives, created by bonus depreciation, to adopt less-than-worthwhile automation.”

In the midst of this stormy environment, there has been a growing anti-AI sentiment from a mix of consumers, artists, teachers and companies. This has been accentuated by a widespread labor union movement that has spent the past year shoring up its contracts with AI protections. Here are just a few of them:

• The Hollywood Unions: This is perhaps the most well-known example of a union win against AI-happy corporations. The Writers Guild of America, Directors Guild of America and Actors’ Guild (SAG-AFTRA) have all secured certain protections against AI replacement. SAG-AFTRA, meanwhile, is currently striking video game companies in an effort to apply these same protections to video game performers. 

• Microsoft and AFL-CIO: At the end of 2023, Microsoft and AFL-CIO (the largest federation of unions in the U.S.) announced a deal that would upskill labor leaders in AI while separately making it easier for Microsoft workers to unionize. At the same time, Microsoft agreed to a union contract with its ZeniMax game studio that included language ensuring AI protections. 

• Financial Times: Last year, union members at FT Specialist, an editorial division of the FT, secured a contract with language that ensures the company will discuss the introduction of new technology before implementation. The union will also “have the right to bargain expeditiously over the effects of the changes.”

• Ziff Davis: More recently, journalists at Lifehacker, Mashable and PCMag secured a new contract with parent company Ziff Davis that includes some specific protections against AI, including the following: “no members will be laid off or see a decrease in their base salary as a result of the implementation of Generative AI technology.”

We talked about this first with Helen Toner, but I think today it bears repeating: how do we want this technology to impact us?

It’s a question everyone should think about.

From that starting point, we can think about how we can make sure that impact is as positive as possible, ensuring a baseline of consent, compensation and transparency.

Without those three elements, it will be used to exploit people.

🤔 Your thought process:

Selected Image 1 (Left):

• “Image 2 handlebars looked clean and regular. Only a real photo could be as blurred as Image 1 and the arm/hand was too complex not to be real.”

Selected Image 2 (Right):

• “You definitely tried to trick me with the wonky-looking hand in the real picture.” (Guilty as charged).

💭 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 cops writing their reports with GenAI:

Close to 40% of you don’t like the idea. A quarter of you said it might be okay if there was explainability and transparency in the systems.

16% said it sounds like a great time saver.

Just give me some transparency:

• “The officers are always going to write reports through the subjective lens of their own memory and experience. Chances are that the AI’s modeling of what happened based on inputs it received is going to be actually closer to the ‘truth.’ Plus, the officers will still be reviewing and filling in blanks.”

Time Saver:

• “Presumably a human would have to review and sign off on the AI-generated report so I’m not sure what the big deal is? It’s ultimately the same as it’s always been and maybe even more accurate than our very fallible human memory.”