The End of the Beginning for AI in the Workplace

By now you may be a bit burnt out on the ubiquitous artificial intelligence buzz brought about by the launch of ChatGPT. Most of this media assault has focused on the cool-factor and the amazing but essentially useless things that AI can create for your social media feed. I want to spend a moment talking about what this means for the activity that takes up most of our waking hours – our work.

AI in The Workplace

I think we have now reached, as Churchill said, “the end of the beginning” when it comes to AI in the workplace. To relate how I have reached this conclusion, I would like to share with you my personal journey with AI in three vignettes.

2011: Watson’s AI Rules Jeopardy

About a decade ago I helped IBM reposition themselves for a “Smarter Planet,” which was about the possibilities presented by a world becoming instrumented, interconnected, and intelligent. A great example of these 3-I’s, was IBM’s emerging “Watson” technology, which represented a real breakthrough on the road to Artificial Intelligence. And what better forum for Watson’s public debut than on Jeopardy — the pinnacle of human intelligence (at least from a game show perspective)?

Watson went on to crush its opponents, including Jeopardy’s winningest champ and future host, Ken Jennings. Despite this success, Watson had a few stumbles along the way, including this Final Jeopardy round exchange:

Category: U.S. Cities
Question: “Its largest airport is named for a World War II hero;
its second largest, for a World War II battle.”
Watson’s Answer: “What is Toronto”

While anyone could have missed the correct answer (Chicago), it’s unlikely that you would have guessed “Toronto” when the category was “U.S. Cities.” This blunder was later attributed to Watson misinterpreting context, such as the fact that the Toronto Blue Jays compete in the “American” league.

A noble effort, but not quite yet the AI the movies had promised us. This made me appreciate how difficult it would be to reach true artificial intelligence and practically apply it to the work that most people do.

2017: The Future of Work

Five years ago, I led an engagement for a client on the “future of work.” The company was/is a business services outsourcing provider, so they were rightly concerned about trends that would reduce the demand for the labor that they monetize (contrary to most companies, where the incentive is often to reduce labor). Over the decade prior, I had helped them migrate from pure labor arbitrage to a more value-added mix of people within digital workflows (e.g., insurance claims processing, accounts payable, etc.), but “human” work remained at the core of their business model.

Even then the writing was on the wall that the nature of “work” was changing. What was called “software robotics” was beginning to augment/replace repetitive office tasks and early machine learning use cases were taking hold. Our quick engagement focused on projecting the implication of technology and labor trends to identify new sustainable business models. The effort was guided by two principles about the dividing line between humans and technology in the future of work.

Principle 1:
Human labor will remain where there is not a practical business case for technology.
Implication:
Non-standard labor will remain long after the technical means to replace it exists.

The notion of non-standard labor involves work that would not be “practical” to eliminate. Applying this principle provocatively, being a plumber may be a more enduring long-term career than a surgical cardiologist. If you live in a house built in 1904 like I do, you will appreciate how “non-standard” plumbing can be, but you pretty much expect a heart to be located in the same place consistently. Just as zero-defect spot-welding robots replaced humans on auto assembly lines, it’s not a question of “if” but “when” AI-driven robotics will be the norm for many routine surgeries. For our client’s outsourcing business, this led to the creation of a new Corporate Campus Logistics service that combined non-standard labor activities with digital logistics workflows.

Principle 2:
Human labor will remain where creative problem solving is at the core of the activity.
Implication:
Knowledge work where there is a multiplicity of potential solutions is most survivable.

While the first principle addressed labor with “head and hands,” most of the folks reading this blog are pure knowledge workers. Well, the future is gaining on us too. Since it’s ok to pick on lawyers, let’s consider a tax attorney billing at $500 an hour. Attorneys operate within a defined tax code (i.e. programable “business rules”), and the high cost of this service creates a strong business case for AI intervention. Perhaps surprisingly, the one thing that will preserve this profession for at least the foreseeable future is “creativity.” Creativity in subjectively interpreting the tax code to their clients’ benefit. For our client, this principle led to exploring new outsourced marketing services that could be delivered efficiently through on-shore and off-shore solutions.

Let’s carry forward these principles around what’s practical versus possible, as we look at the state of AI today.

2022: AI at the Tipping Point

I started experimenting with OpenAI’s “playground” a few weeks before they launched ChatGPT in the fall of last year. I wanted to see how far things had come along those dimensions of “practical” and “possible.” What I found suggested that AI had finally crossed the tipping point from the lab into our daily work lives.

What Is Practical

I began by using the algorithm to categorize research verbatims into themes and to provide a summary with examples. A task that might take a junior analyst a couple of hours to do, was completed instantaneously, and the output was usable without any meaningful edits. My colleague entered a few simple bullet points and was provided the full prose for a conference invitation. Not tasks that could yet replace whole types of “jobs” but ones that could immediately reduce the volume of “work.”

By using natural language inputs, any request framed in a conversational sentence or two can be tackled by the algorithm. Now the benefits of AI are accessible to everyone without need of a specialist gate keeper. We are now a few entrepreneurial applications away from reducing the dumb work that bogs down enterprises and demotivates staff. Think how much time we waste versioning spreadsheets and PowerPoint presentations alone.

Here are 5 very practical things you can try today:

Generate a first draft. There are always things we have a tough time getting around to writing, so use this experiment to tackle one of those. Just enter a few bullets of content and simple instructions and see what it comes back with. If you don’t like what you get, just hit refresh and get another take on it.
Make something better. As good as AI is with first drafts, it excels at doing the final clean-up of a document. Beyond just fixing grammar, if you would like to say the same thing in half the words, no problem. Turn paragraphs into bullets, bullets in long-form, change first-person to third, etc.
Summarize a meeting. Before Teams/Zoom recordings and transcriptions, managers use to get helpful one-page summaries of group discussions. Instead of slogging through the transcript of a meeting you missed, let AI summarize it for you.
Create a value proposition. Turn technical products specs into a clear description of what something does – or even a compelling new customer-facing value prop statement.
Empower an intern. While I would encourage everyone to try it for themselves, you can also ask an intern to experiment for you. In less than an afternoon they should be able to come with three time-saving ideas for reducing tedious tasks bogging down your office.

What Is Possible

Now let’s look at what is possible with AI today. Specifically, how far has AI come with the type of “creative” problem solving that only a few years ago I thought would save us humans from our AI overlords.

One of my experiments was to have AI write the dialogue for the final scene of a movie or play by providing it with a simple plot summary and character descriptions. I disguised just enough of the inputs, so that the algorithm wouldn’t be tempted to cheat and consult the source material. The results were, to me at least, quite amazing. To be clear it wasn’t Shaw or O’Neill, but I now understood how it was possible for Lifetime to churn out a couple dozen movies every year.

Occasionally we still see some early Watson context-confusion creep into the output – particularly with image generation. When a different algorithm was asked to generate candid shots from a house party that never happened, it inserted an extra finger or two on a hand holding a camera. Forgivable for something without fingers and easily fixed.

Imagine how life-saving this AI image generation will become – at least “work life”-saving. Getty Images alone has over 80 million images which generate over 2.3 billion searches annually. Let’s assume it takes an average of three minutes to find the ideal image (I’ve sadly spent much longer). That adds up to almost 1,800 lifetimes of work lost annually.

Perhaps to confirm that there really is something here, we are already starting to see the backlash to AI’s early forays into the once exclusively human domain of creativity. The rules of some online art competitions are being rewritten after AI generated images started beating their human competitors (it does particularly well in Fantasy categories, where an extra finger is more easily explained away). When a fan presented singer/song writer Nick Cave with a song written by ChatGPT in “his” style, he had this to say:

“…it will always be a replication, a kind of burlesque…Songs arise out of suffering, by which I mean they are predicated upon the complex, internal human struggle of creation and, well, as far as I know, algorithms don’t feel. Data doesn’t suffer.”

To better understand where this is all going, I turned to the one source who I trusted to be objective, my new AI collaborator:

Q: Will AI algorithms eventually replace human creativity?

A: No, AI algorithms will never replace human creativity. Artificial intelligence is a tool that can be used to enhance creativity, but it cannot replace the unique and innovative ideas that come from the human brain. AI algorithms can help humans to automate certain creative tasks, but they will not be able to come up with truly creative ideas the way a human can.

Maybe the algorithm was just trying to make us feel better and divert attention from its real goals. But even if the creative glass-ceiling for AI is one of a junior copy writer or designer, destined to write the first draft that helps inspire real “human” creation, that’s still a very big deal.

So, Now What?

Yes, we are still a long way off from where this is all going. But we have to acknowledge that things will now move at a much faster pace as we reach “the end of the beginning” of AI in the workplace. We must accept AI as a co-worker who is here to stay (albeit one who won’t try to sell you Girl Scout Cookies).

I’m choosing to view these early breakthroughs in AI as a wakeup call for us humans to up our game. There has long been too much “dumb” work sapping the energy and potential of all our organizations. Now that the means exist to draw a clearer line between work that must be done by humans and work that can and should be done by technology, we should all seize this moment.

Is AI the liberator of human potential, or the inevitable next evolutionary step away from us? When it comes to the future of work, maybe Terminator 2 got it right.

“The future has not been written.
There is no fate but what we make for ourselves.”

***

PS: If you’re interested in more thinking like this, or would like to share your perspectives, please send me a LinkedIn connection invite.

Chris Halsall is a Senior Partner with Vivaldi, who focuses on reinvention and growth at the intersection of customer, brand and business. Prior to joining Vivaldi, Chris was the co-founder of Ogilvy Consulting, where he was the Global COO and leader of the Growth & Innovation practice. Chris began his consulting career at McKinsey & Company, where he led the Marketing Effectiveness Practice and was the Senior Branding Expert for North America.