Top 10 Stories of the Year: It's All About Generative AI

Generative AI was the hands-down winner in this year’s top 10 stories on AI Business, capturing nine out of the 10 most read. For good reason: generative AI has ushered in a new way for humans and machines to communicate.

Among the top 10, four delved into ChatGPT’s new or optimized capabilities. OpenAI’s revolutionary chatbot turned one year old this past November and it is growing up: it has new abilities that are powered by stronger GPT language models.

Without further ado, here are our most popular stories in 2023:

#1 Meet Caryn, Your Generative AI Girlfriend

Published: June 1

AI startup Forever Voices created a generative AI girlfriend after the influencer, Caryn Marjorie. What made this story so popular among readers is that the avatar, also called ‘Caryn,’ is no ordinary digital friend.

Generative AI enabled digital ‘Caryn’ to interact dynamically with users, coming up with fresh new responses instead of canned replies. Caryn returns impressively human-like responses to user inputs, and even generates audio.

Forever Voices has also made clones of Steve Jobs and Taylor Swift.

Read the full story

Credit: Caryn Marjorie on Snapchat

#2 OpenAI’s Code Interpreter Lets ChatGPT Play Data Scientist

Published: July 10

The most popular ChatGPT-related story on AI Business this year, the launch of Code Interpreter marked the beginning of a huge shift for OpenAI’s chatbot.

Instead of merely generating code, Code Interpreter let ChatGPT users execute code right there in the chatbot. Users could perform detailed data analytics and visualization without having to write a snippet of code.

Code Interpreter marked the beginning of a series of ChatGPT updates, with voice inputs and image generation later following.

Read the full story

#3 OpenAI’s Six Steps to Improving Your Prompts to Get Better Results

Published: Dec. 18

A come-from-behind article, OpenAI’s tips on how to write better prompts resonated strongly with readers this year despite being just recently published.

The startup quietly published a prompt engineering guide under its website’s documentation section, sharing tactics and tips for users to get better results from large language models like GPT-4.

OpenAI offers six steps for optimizing prompt-writing, noting that some of the methods can be combined “for greater effect.” Users can also explore various prompt examples to get the best out of their inputs.

Read the full story

#4 ChatGPT Can Now Give You Real-time Information

Published: Sept. 28

Two months after Code Interpreter dropped, OpenAI supercharged ChatGPT by giving it limited access to real-time information via Bing search.

Prior to this update, ChatGPT’s data range ended in September 2021. By giving the chatbot internet access, ChatGPT was able to provide current and “authoritative” responses, complete with direct links to sources.

This feature was only available in the premium version of ChatGPT.

Read the full story

#5 7 AI Programming Languages You Need to Know

Published: May 5

As the most popular evergreen listicle from AI Business this year, the seven most commonly used programming languages in AI include the popular Python and Java that any AI developer needs to master.

Entries like Julia and Haskell were included to inspire learners to pick up lesser-known languages and get a leg up on competitors.

Read the full list

#6 Meet Alpaca: The Open Source ChatGPT Made for Less Than $600

Published: Mar. 20

Among the first open source large language models built atop Meta’s LLaMA model, Alpaca caused quite the commotion when it launched in late March.

Costing just $600 to develop, users scrambled to get their hands on it. And Alpaca performed well – until its public-facing demo debut. Stanford researchers had to pull the plug due to the model’s hallucinations. The team blamed “inadequacies of our content filters.”

Despite Alpaca’s ill-fated demo, the model marked the beginning of a swathe of open source AI models that launched in 2023 - many of whom had Meta and its LLaMA to thank.

Read the full story

#7 Goldman Sachs: Generative AI Could Replace 300 Million Jobs

Published: Mar. 29

Despite impressive advances in AI in 2023, fears around job losses loomed large. Reports like this one from Goldman Sachs did not exactly help to quash concerns - its economists said that generative AI could replace up to one-fourth of current jobs globally, or 300 million.

About two-thirds of jobs in the U.S. and Europe could be exposed to some degree of AI automation. Jobs most at risk in the U.S. were office and administrative support positions, followed by roles in law, architecture and engineering. The least likely to be affected are manual jobs such as those in maintenance, repair work and construction, among others.

Read the full story

#8 Create Songs with AI: DeepMind's Lyria for YouTube

Published: Nov. 16

AI Business covered a lot of fun and exciting models this year, but Lyria from Google DeepMind was one our readers explored the most.

Built to power music generation features on YouTube, Lyria can generate high-quality music – including instrumentals and vocals.

Read the full story

#9 WAICF '23: Yann LeCun Sets Us Straight on Generative AI

Published: Feb. 10

It has been quite a year for Meta Chief AI Scientist Yann LeCun. He led a team pushing the boundaries of open source AI and was made a Chevalier (Knight) of the French Legion of Honour by President Emmanuel Macron.

At the World Artificial Intelligence Cannes Festival (WAICF), he was a standout speaker. So many people wanted to hear him speak that even the overflow room requiring an overflow.

More than LeCun’s presence was his actual description of ChatGPT that captivated reader interest: He said it was little more than a “typing aid.” The story went viral after LeCun shared AI Business’ coverage of his speech on LinkedIn.

Read the full story

#10 ChatGPT Can Now Summarize, Analyze Uploaded Documents

Published: Oct. 30

OpenAI gave ChatGPT the ability to look through and analyze longer documents, available to users of its premium subscription service or enterprise version. Users can interact with the chatbot to ask questions about the document.

For example, you can upload a research paper on machine learning and ask ChatGPT to summarize it in simple terms. Or input sales reports and ask them to generate potential trends. Or even use the vision feature to take a picture of an object and use that image to influence DALL-E 3 generations.