Introduction
I have always been curious about technology.
When I was 15, I spent the summer in the UK and had a job working in the customer service department of the company my father worked for (thank you, nepotism). The company made portable meter readers for utility companies. The utility workers entered a meter reading into the machine at the customer’s home and then printed a bill and left it with the customer on the spot. In 1985, this was cutting edge. The innovation at the time was a portable memory technology that allowed the prior readings and other data to be stored in the machine and accessed remotely. For those of us who are old enough to remember, there was no wifi in 1985, so it was a pretty cool device. Although it was a bit heavy, and not very reliable – as I found out working in customer service.
I didn’t continue my IT career after that summer job. The following summer, I worked at a law firm where the most exciting technology was the fax machine. Over the following decades, my curiosity about technology was more as a consumer of new software and devices as they became available. I didn’t really think too much about how the technology worked other than having a vague understanding that most of the apps on my smartphone are algorithms. That changed when ChatGPT was made available to the public in 2022. I found myself asking what exactly is a large language model? I was concerned when brilliant scientists started raising the alarm that the technology needs to be regulated before it is “too late.” I wanted to know more.
My curiosity led me to sign up for Oxford Artificial Intelligence Programme. The course provides a conceptual understanding of AI technologies, including machine learning, deep learning, artificial neural networks, and algorithms. It also focuses on the social and ethical implications of AI. I had to do homework for the first time in a very long time. The course ended last week, and I was not disappointed in how much I learned and the practical knowledge I will be able to use in the practice of law. Not to mention that I will soon be able to joke that I am a “graduate” of Oxford University.
Over the next few editions of our newsletter, I will write about some of the interesting things I have learned about AI that I think will be of interest to HR professionals. In this newsletter, I will start by asking whether AI will replace employees.
AI Replacing Employees
It is not a new concept that machines can replace workers. This has been happening incrementally since the first industrial revolution. However, many believe that the recent advances in AI are different, and we are about to enter the 4th industrial revolution. Yes, there have already been three industrial revolutions. The first revolved around the steam engine. The second is around the internal combustion engine and electricity. The third started in the 1960s with the advent of electronics, including telecommunications and the computer. There is some debate if we are entering a fourth (Utko, 2018), but “Industry 4.0” is described as the exponential expansion in technology that is uprooting industries.
I have learned that there is no single definition of AI, but one I read recently is “[t]he art of creating machines that perform functions that require intelligence when performed by people” (Kurzweil 1990). I’m pretty certain that the meter reading machine I described above replaced a few clerks whose job it was to collect the meter readings from the workers in the community, calculate the amounts owing, and send out the invoices. That is not the kind of AI that brings in a fourth industrial revolution, but innovations such as ChatGPT just might.
Large language models like GPT-4 are set to change the job landscape because they can process and generate human-like text, automate complex tasks, and provide insights derived from large datasets. Automation of low-skill and low-income jobs is anticipated. However, jobs that require high social intelligence, creativity or perception are considered low risk currently. Some experts believe that AI is not expected to make large parts of the white-collar workforce redundant but to augment their abilities through the tools it creates. This may increase employment due to its contribution to productivity (Vander Ark, 2018).
Here are a few current examples of large language models that we interact with every day:
- Chatbots and Virtual Assistants: Many companies have integrated sophisticated chatbots powered by large language models into customer service operations to provide real-time assistance to customers, offer support, answer frequently asked questions, and guide users through troubleshooting processes.
- Content Generation and Curation: In the media and content creation industries, language models like GPT-3 are used to generate articles, marketing copy, and even creative writing. They assist human writers by providing initial drafts, creative prompts, or by automating updates to factual content.
- Code Generation and Bug Fixing: Programming assistants like GitHub’s Copilot, which is powered by OpenAI’s Codex, leverage language models to understand natural language requests and generate coding scripts, debug codes, or even provide code suggestions as a programmer types, thus improving developer productivity.
- Business Intelligence and Data Analysis: Large language models can analyze large volumes of text data to identify trends, generate reports, and provide actionable insights, helping businesses to make data-driven decisions.
- Legal Document Review and Analysis: Language models are increasingly used in the legal field to review documents for discovery, draft and analyze contracts, and provide legal research assistance.
What is apparent from these examples is that despite the extraordinary capabilities of large language models, only the first example above involves robots taking over human work. Frankly, from my personal experience with chatbots, they are a poor substitute for speaking to a live agent. So, overall, at this time, it is safe to say that the robots are here to serve us and definitely are not taking over.
In my practice, I am currently piloting a Thomson Reuters Product—casetext CoCounsel—which advertises that it does “shockingly thorough, accurate, and efficient work.” It reviews documents, creates timelines from documents, summarizes documents, and will integrate with the legal research platform Westlaw. I intend to make my own assessment of the tool’s accuracy and efficiency. As a side note, the list of five examples above was generated using casetext, so it did save me some time there.
Stay tuned for the next article, in which I discuss developing an ethical framework for implementing AI in the workplace.
Resources:
Kurzweil, R. 1990. The age of intelligent machines. USA: MIT Press.
Utku, A., The fourth industrial revolution is a hoax and it has never happened.: https://medium.com/@alperutku/the-fourth-industrial-revolution-is-a-hoax-and-it-has-never-happened-fed1a7d9f283
Vander Ark, T., The Rise of AI: What’s Happening, What it Means, How to Prepare: https://www.gettingsmart.com/2018/03/13/rise-ai-whats-happening-means-prepare/