Welcome to the first edition of The Path by Pathfindr.ai. Every week we will give you honest, relatable insight on the practical opportunity and impact of AI on businesses.
It’s no doubt a massive opportunity, but most struggle to think of more than one or two practical ways to use it in their business, and for many their thinking is limited to ChatGPT.
Today’s newsletter is all about the implications for the corporate practitioner, sometimes called a knowledge worker or what we once commonly labelled white-collar workers. The attitude towards AI ranges from dismissive and sceptical through to angst on whether they’ll have a job. Today’s article will share my honest current thoughts on this topic.
What I won’t be addressing is a macroeconomic view, for instance the net effect of job destruction and job creation and what history tells us. Benedict Evan’s article on AI and the Automation of work is an excellent view of this.
I use the term ‘practitioner’ to highlight these people serve instead of supervise. They serve customers external or internal (in an air-conditioned office with easy access to coffee, corporate biscuits with long shelf life and/or savoury crackers with similar longevity properties).
This is the lawyer, the accountant, the call centre agent, the HR partner or the sales rep. The architect, teacher, financial controller or software engineer. AI has a role to play with supervisors or management, but that’s a topic for another newsletter.
Before we can answer if AI is coming for your job, let’s baseline what AI is good at, and think of this in the form of activities, not jobs. The second step is to measure the impact as eliminating, improving or amplifying the activity (AI actually adds work in certain cases, but has an impact in other ways).
In other words, does it get rid of that activity, make it easier for you, or make you able to do things you couldn’t before (think of it as giving you superpowers, like having the ability to use Google Maps to find out how to get anywhere, anytime, in any city).
If your job is entirely made up of activities that can be eliminated by AI, yes, you have some thinking to do. But you might be on the opposite side where AI allows you to amplify your activities (eg. A marketer that can address 200 segments instead of 10, or an English support agent who can provide chat in Spanish).
In the middle is ‘improve’, like giving some easier access to information that saves you a bit of time, or makes it easier for less experienced people to do a similar task. AI assistant chatbots like ChatGPT is a great example.
Another important concept is what you are measuring. If you run a call centre that takes a million calls a year, but the hold time is 1 hour per call and AI helps your agents do double the amount of calls, you can either have customers wait for 30 minutes or you can reduce your call centre headcount. If you’re about to go bankrupt you’ll decide one way, or your brand isn’t giving the service customers are paying for you’ll decide another.
If you’re in sales you might measure how many deals you close, or what % of deals you lose at a certain stage in the sales process, but the goal is not to get less sales reps, it’s to increase sales.
An overlooked critical concept of AI in our jobs is how much it’s embedded into our existing workflow. Think of it as integrated or optional. ChatGPT is an optional tool, if you give it to 10 people to help with their job, some will use it regularly, most will use it infrequently, and some will never use it.
Some will rave about it, others couldn’t be less interested, and maybe even despite what they say, their measurable performance has little relationship with AI.
However, if you make AI integrated into a workflow, it’ll be used every time and the benefits and experience can be measured and improved.
For instance embedding an ‘AI contract review’ that a senior lawyer requires their junior staff to use before asking for a final review.
The intention would be that the senior lawyer has directed the AI in a way to find the 95% of trivial things that they pick up every time. If the lawyer continually finds common issues that the AI tool misses, she can instruct it how to improve, or bin it and replace it with another one.
If someone presents the senior lawyer with a contract, she can immediately ask “have you used the AI contract tool to check for common issues”. Integrated doesn’t mean it needs to be a feature in existing software (although that can help), rather just as a requirement in a workflow process.
So if we think in terms of AI impacting activities and workflows, with the lens that it might be integrated or optional, let’s dive into what AI is good at.
Firstly, it’s important to acknowledge that AI is a broad term, it’s a concept, and it’s ridiculous for me to start a sentence like “what is AI good at” as it’s the same as saying “what is transportation good at”, but I’m going to run with it anyways.
Right now, AI is very good at:
When you see the magic happening with an instruction like “Can you create me a briefing of all the South African rugby team matches over the last 10 years in less than a page easy for an 8-year-old to understand, highlighting milestones, heroes and upsets”, what’s happening is a combination of all of the above. It’s identifying what you’re asking for, it’s then performing a thorough search, it’s then summarisation its findings, and finally translating it into a style suited for my son.
Before understanding what AI isn’t good at, we should acknowledge that AI is just reflecting patterns in its underlying training data. It’s a mind-trip to simplify it like that, but it’s accurate.
One could argue that that’s exactly what people do - reflect their training data - but there are emotions, instinct, millions of micro-signals that we are not even aware of developed through millions of years of evolution, and even more strangely, if there is a concept most people understand, but there’s no word for it and it’s not commonly discussed, AI will have no knowledge of it.
As such, anything that is influenced by human emotions, nuance, emotional resonance, cultural norms, instinct, and fields where subtle complex variables impact the outcome AI will have a limited impact. Thankfully, work that has these components is usually what we find more fulfilling.
Essentially true creativity, complex problem solving and innovation is still very difficult for AI. AI might help you understand a customer type, but it won’t help you deeply understand that customer. It might help you start a good conversation with them, but it won’t be able to predict how the conversation will go or even how you should respond.
We barely understand why we behave in certain ways, AI is no better, but it will have a guess based on what’s in the training data.
So it may help you craft a message to help sell something, but it will need your help to make it emotionally compelling. As Chris Voss states in “Never Split The Difference”, the emotional side of negotiation, how and when you say things as well as subtle behaviour is far more important than what you say.
Subtlety and nuance actions do not have subtle or nuanced effects.
So I still strongly suggest that most important aspects of fields like negotiation, sales, marketing, psychology, leadership, mentorship, entrepreneurship, architecture, software engineering and customer service are things that AI is not particularly good at (but can significantly help in lower-value work in these fields).
Finally, there are many activities that AI can do a fantastic job at replacing junk work.
AI should replace the journalist who regurgitates 3 articles that others have published and brandishes as their own.
AI should replace the insurance broker that forces you to read the fine print yourself instead of diligently explaining it to you (or else why not just do it online for cheaper).
AI should replace the consultant who just takes what the client has said, pretties it up, puts it on a PowerPoint slide and presents it back and charges a fee instead of actually performing emotional labour by crafting insight and engaging in challenging discussions.
The articles and slides exist, look great, and probably impress some people, but they’re junk. AI should disrupt this kind of work, if you’re going to create junk, charge a fee for value but deliver none, just get AI to do it and pay a $20 ChatGPT plus subscription instead.
The short answer is - AI is going to appear in your job but what impact it has is - ‘it depends’. Some of it might be invisible, you might decide to use it if you want to, it may be a huge amplifier, or it may be a requirement. But no doubt it’s going to appear in your job in some form.
Personally I am an AI optimist. I think of the graduate with English as a second language preparing for their first job interview, or those with learning difficulties being able to have a 24x7 infinitely patient teacher. A junior software engineer able to be more self-sufficient instead of interrupting the time-pressed senior architect.
Giving small businesses the capability of large design agencies, photographers, and graphic designers that they otherwise wouldn’t have been able to afford, giving them an impressive e-commerce website that faces up to any large multinational.
But maybe I’m just naive. What I do know though is that starting with understanding and exploration without fear should be the first step.
See you next week.
Dawid
.jpeg)
