This AI job is paying $300k per year! Plus more AI news

The December 20 intelligist AI newsletter

Greetings, intelligists!

Today’s newsletter starts with some rather funny news. You might have heard about the viral incident related to a Chevy dealer.

It is a tale of how not to integrate AI into your business. It is also a tale of how there will be many, many predators on the prowl looking for loopholes as more and more businesses roll out AI in their operations.

So take it with a laugh but also as a cautionary tale. Here it is:

A Chevrolet dealership in California found out the wrong way what happens when you carelessly implement AI.

The car dealership implemented a ChatGPT-powered customer service chatbot on its website, and the management probably felt quite cutting-edge for doing so. The idea was that the chatbot could help customers decide on the best model for them, as well as make appointments and answer general inquiries.

Image generated by the author in Dall-E 3

However, smart-alecky users started messing with the chatbot to see what it could do. One troublemaker had the chatbot help them create complex computer code. Another convinced the bot to recommend a Tesla over the in-house GM models. Finally, the best of them was a user who found a way to manipulate the bot into selling him a 2024 Tahoe for only $1 in a "legally binding agreement." The wiseacres effectively tricked ChatGPT into giving away a car.

All in all, this incident could have quickly gotten out of hand. Complex chat requests and answers could run up a hefty bill for the site owner. And while the agitator is unlikely to get a nearly $60k Tahoe for a dollar, this incident brings up some other important points. For example, if you are going to implement AI features into a website, you need to make sure that whoever is doing it knows what they are doing. This means not just making a chatbot work but understanding all other consequences that it could imply.

On the one hand, this means you need somebody with strong technical knowledge to actually do implementations. But it also suggests the importance of having managers or officers who have a high degree of understanding of AI as well as organizational operations. For example, the role of the AI officer is becoming more important. This role includes understanding both the technical aspects of AI and how AI can impact operations, the business model, human resources, and other aspects critical to the organization.

The need for having qualified AI officers and managers will surely become even more important in 2024, with AI-related jobs slated to grow exponentially.

AI training data includes child abuse images, Stanford researchers find

In today's disturbing AI news, Stanford researchers have found that child sex abuse images have leaked into AI training datasets. Specifically, this illegal material, known as CSAM, made its way into the LAION-5B model, a reference image database. The researchers asserted that the appearance of such images could be related to similar images that are filtered through search engines and social networks.

Often, image and text training datasets for AI are scraped from these types of sites. While the appearance of these images might be unintentional, it points to some key concerns. Many AI models scrape data from all over the web, and it is evident that sometimes, this data can be "bad," incorrect, and even illegal.

This hot new tech job pays $300K a year — what's an AI prompt engineer?

The hottest new job in tech isn’t “forward-thinking futurist CEO” or obscure computer language programmer.

Instead, it is that of “prompt engineer.”

The role of an artificial intelligence prompt engineer, responsible for formulating questions and instructions to obtain refined answers and images from AI programs like ChatGPT, is in high demand and offers a substantial salary, with some positions reportedly paying over $300,000.

This rapidly growing sector is attracting professionals with strong writing and AI usage skills. It’s a job that requires thinking like a user to craft effective prompts and thinking like a GPT to find ways of manipulating it into creating good responses.

Meanwhile, industry leaders caution that the rapid advancement of AI technology may displace certain white-collar workers, including coders, computer programmers, journalists, software engineers, data analysts, paralegals, and legal assistants.

As I always say, technology gives, but it also takes away. I have the feeling that we will suddenly see a tide of prompt engineers prompting out of the woodwork on LinkedIn. Beware: not all will have much of an idea of prompting, engineering, or even AI!

UK Supreme Court rules that AI cannot be a patent inventor

The UK's Guardian newspaper reported that the UK Supreme Court made a landmark ruling Wednesday. The high court ruled that AI cannot be legally named as the inventor of a potential patent, after a US technologist attempted to register patents with his AI as the creator.

This ruling might seem trivial, but it is a great blow to potential inventors who happen to be AI robots. Hey, wait a minute…

That concludes today’s intelligist. Have a great Wednesday, and tune in tomorrow!

Joaquin