The most in-demand AI skills — and how companies want to use them

Companies are looking to hire more people with experience with artificial intelligence (AI) as the arrival of generative AI tools and platforms forces enterprises to re-think their digital strategies, according to new studies from employment websites.

The studies also documented the AI skills most sought by organizations who hope to use the technology to boost efficiency and production by augmenting and/or automating employee tasks.

Job listing site Upwork on Tuesday released its study on the skillsets of job seekers accepting jobs as well as the search and hiring behaviors of companies in relation to fastest-growing skills category: generative AI. It found that 49% of hiring managers plan to hire more independent talent and 49% say they will hire more full-time employees — both a result of AI deployment plans.

AI was the fastest-growing category on Upwork in the first half of 2023, with generative AI job posts up more than 1000% in the second quarter of this year compared to the end of 2022; related searches increased more than 1500% in the same time period.

The top 10 Generative AI-related skills listed in Upwork job seeker profiles with the largest quarter-over-quarter growth (Q2 2023 vs Q1 2023) were: Large Language Model (LLM); Generative AI; You Only Look Once (YOLO); Object Detection; Stable Diffusion; Prompt Engineering; ChatGPT; Azure OpenAI; AI Chatbot; and AI Text-to-Speech.

The top 10 generative AI-related job hires were: ChatGPT, Natural Language Processing (NLP), Tensorflow, Image Processing, Pytorch, AI Content Creation, Midjourney, AI Chatbot, LLM tuning, and Stable Diffusion.

Upwork determined the top AI hires by looking at each job posting and the skills freelance job seekers attached to their online profile; that included gen AI tools, like ChatGPT.

“We looked at the number of jobs hired during January 1 – June 30, 2023 that included Generative AI skills,” said Kelly Monahan, managing director of Upwork’s Research Institute, which performed the study. “We reported on the top 10 Generative AI jobs hired that included one of the Generative AI skills listed in the job posting. Chat GPT rose to the top over that time period. The job hire itself could be a range of activities such as hiring an AI developer who can work with Chat GPT, a Chat GPT writer or a Chat GPT expert to consult on use cases for a company as examples.”

The fastest-growing generative AI-related searches between Q1 and Q2 of this year, according to Upwork, were:

1. AI Content Creation
2. Gradio
3. Azure OpenAI
4. Convolutional Neural Network
5. Large Language Models (LLMs)
6. Generative AI
7. AI Chatbot
8. Midjourney
9. Prompt Engineering
10. Pytorch

The sentiment around hiring and the surge of activity from companies interested in gen AI are strong early indications that businesses are paying significant attention to the opportunities the fast-evolving technology can provide, according to Upwork. Companies are shifting their focus from singular gen AI tools toward applications and services, including AI content creation, Azure OpenAI, and prompt engineering.

“Generative AI is following a similar adoption curve to other new and emerging technologies,” said Monahan. “When it is first broadly released to the public, people are still determining the specific use cases of the technology, so their searches begin at a broad level. In the case of Generative AI, ChatGPT was the first tool that sparked public attention. As a result, our platform data shows that many of the searches at the beginning of the year were from people coming and looking for freelancers with ChatGPT skills.”

In fact, ChatGPT commanded the largest volume of searches by companies in the first half of 2023. And ChatGPT’s creator, OpenAI, continues to see “a first-mover advantage” with its tool, including the underlying large language models — GPT-3 and GPT-4. ChatGPT is at the top of the list of generative AI-related search terms from those looking to hire, followed by Google’s BERT and Stable AI’s Stable Diffusion.

On social network and professional networking site LinkedIn, the share of job postings mentioning GPT or ChatGPT increased nearly six-fold from May 2022 to May 2023. And the number of LinkedIn members who hold, or have been holding, “head of AI” positions has nearly tripled in the last five years.

As people have learned more about the capabilities of gen AI, they are beginning to develop a deeper understanding of how it can actually be used. This maturity cycle, Monahan said, is reflected in gen AI-related searches. “Instead of looking for just a single tool, like ChatGPT, they are searching for the diverse use cases of generative AI technologies like AI content creation, services like Gradio, used for building machine learning web apps, and prompt engineering,” Monahan said.

The changes in the hiring market reveal a deeper recognition of the skills needed to implement AI to solve business problems, such as working with Large Language Models (LLMs), the role of prompt engineering, and object detection.

Object detection is a computer vision technique in which models can be trained to locate, identify, and analyze objects in a digital image or video, according to Monahan. You Only Look Once (YOLO) is a specific object detection tool Upwork sees growing quickly on freelancer profiles as a top generative skill in their profiles.  

In an earlier July study, Upwork found the number of weekly job postings related to genAI like ChatGPT, Google’s Bard, and Cohere’s Coral, were up more than 450% from a year ago.

The share of global English-language job postings mentioning GPT or ChatGPT is up 21-fold since November, according to LinkedIn’s Future of Work report, which was released last week.

In fact, 47% of US executives believe using gen AI will increase productivity. Additionally, 44% of executives plan to increase their use of AI at their organization in the next year, and 40% think the technology will help unlock more growth and revenue.

“The rise of AI talent and members adding AI skills has been accelerating since 2016,” LinkedIn said. “With the launch of ChatGPT, we are seeing the promise of AI and how it’s reshaping the skills required at work and at the same time helping people become more productive.”

An analysis of how AI skills are emerging across 25 countries shows that the pace at which LinkedIn members added AI skills to their profiles nearly doubled since the launch of ChatGPT, rising from 7.7% (May–November 2022) to 13% (November 2022–June 2023), LinkedIn said.

While AI has been used for more than 50 years, generative AI — the algorithm’s ability to generate text, images, or other media — has become a highly sought-after capability, according to Erick Brethenoux, Gartner distinguished vice president analyst.

AI can help organizations with decision support, parse thousands of data points or records in seconds, and replace mundane tasks that can free up employees to perform more important, and creative, jobs. “Generative AI was born out of machine learning techniques. People added rule-based systems and natural language processing and knowledge graphs to AI,” Brethenoux said.

For example, gen AI applications can assist in decision augmentation, or the collaboration of humans and machines, Brethenoux said.

“I cannot analyze 7,000 dimensions at once,” he said. “Machines can do that. So, then, great — let them analyze the data and find the patterns. Then how to apply those patterns can be at my discretion and could also be helped by a machine.”

Additionally, LLMs can be tailored or customized with APIs to ingest corporate data or fine-tuned through prompt engineering.

For example, OpenAI’s GPT-4 LLM is pre-trained as next word or content prediction engines; that’s how most businesses use them — out of the box, as it were. Pre-trained LLMs, which can contain billions or even trillions of parameters, work relatively well at feeding mostly accurate and compelling content.

At the very least, pre-trained LLMs can be used as a jumping off point by organizations looking to automate some tasks. But, pre-trained LLMs are often trained using older data and are bloated with information.

As a result, LLMs are poised to shrink, not grow, as vendors seek to customize them for specific uses that don’t need the massive data sets in today’s most popular models. Industry-specific use cases, say for financial services or healthcare, require LLMs to generate more unique outputs using the language of those industries and organizations. That will require workers with skill sets that can customize LLMs for more specific business purposes.

As more employees are trained on AI, fears of job displacement from the technology may subside as workers see the potential for new AI-powered enterprises. For example, there are now 300 AI courses offered on the LinkedIn Learning platform.

“There are certainly online certification courses that professionals can take to update their resumes and be on top of new and emerging generative AI skills,” Upwork’s Monahan said. “There are plenty of places to start, like on Coursera or Udemy, but companies that have their own open AI services, like Microsoft and Azure, also offer online training opportunities.”