This is why AI won’t take our jobs for years and years.

I saw this image this week about jobs under threat from AI, and it made me realize how little people think. Tech won't take jobs, only roles and tasks.

Let me explain more.........

In the 1880’s a simple but transformative idea was introduced into the world of business with a simple name, “A time and motion study”.

Quite simply, people watched precisely how other people did their jobs. Recording with pen and paper, what they moved where, how often, how much did they lift, when did they rest, all with the aim of figuring out one of lifes’ eternal questions; what do people actually do all day.

It’s still used today, but generally only in more complex physical lines of work like Manufacturing, Healthcare, or Warehousing. Which seems a shame because it would be killer to do this with most jobs.

I’ve long been obsessed with how profoundly inefficient almost any company and job in the digital age seems to be. Myself absolutely included.?

Can you imagine someone watching how you spend your time and establishing how much most of it REALLY matters. Can you imagine the same being done to those around you??

I’m not talking about snatched glances of Twitter, nor small talk at the proverbial water cooler, nor the morose scrolling on Instagram on the loo, I’m talking more generally about the time we think we are working well.?

Now, we surely know that inefficiency comes at all levels of work.

Macro level inefficiency

At the very highest altitude, many staff don’t really know what the vision of their company is, they don’t have a strong sense of how to best prioritize, a lot of work these days is well done but rather useless. Did you really need to make a Metaverse strategy, did the innovation day actually make any needles move, you get the idea. How work is created by an Exec firing off an email saying “ Make a report on the Blockchain for Monday” after reading a Wired article moments before take off, only to be cascaded down and outwards to agencies.

Mid level inefficiency

At the day to day level, time is also spent poorly. It may be collating weekly reports which get sent to 100 people, and only one person opens them once per quarter.?Perhaps you’re sending out excel templates to 35 markets, before collating them painstakingly by hand, when a collaborative Google Doc link could save you HOURS. It may be using bad software to do expense reports, duplicating the work of others because you can’t find a badly named file.

Micro level inefficiency

But it’s at the task level where things are even more strange.

Most companies are spit into departments, each with many Jobs (roles) , which are typically split into many Objectives (outcomes, goals or procedures) and each of these are made up of individual Tasks (activities, processes, many of which are split into smaller components or Action Items.

A quantity surveyors Job may be to:

- Manage project finances and contractual relationships at all stages of a building project.

?

The Objectives of this role may be to:

-Manage costs of materials in the projects in the planning stage, prep stage during construction ?

-Prepare, write, manage tender bids and relationships with contractors.

-Manage contract documentation

-Communication: Collaborate with project teams

The Tasks of one objective ( Collaborate with project teams )?may be to:

-Create and share a daily excel of current costs and cash flow

- Use WinQS to estimate construction costs

-Highlight upcoming stages of project work, using Powerpoint, Vector Snape or some CAD program

-Create and share a weekly word document, to act as a status report

-Organize and lead a weekly status with all Contractors and Client

The Action Items for one task ( Organizing Weekly Status) may involve,

- Sending a weekly Calendar invite

- Following up with those who decline it by email

-Sending out Zoom link.

- Sharing excel and Word documents in time for the call.

- Recording time spend on call in Time Management software

- Sending a follow up note with urgent actions in email.

What you can see from this incredibly boring, but simplistic analysis is how even one Task, from one Objective, from one Role, is a mess.

And having spent time now in the worlds of Architecture, Engineering, Advertising and Consulting, it amazes me as to quite how similar almost all jobs are. They all involve near endless toggling between calendars, emails, word files, excel and presentation software.

We have designed the entire world of work, and out lives, around software, we’ve never made software to work around us.

And as such, our roles are all incredibly inefficient. Even the most mundane, simple, non urgent, not vital procedure can take hours and hours.

We’ve never once gone back to basics and rethought the software around us, or built a new operating system for the office, we’ve made Office 365, not “do the job of an account exec” ?

Instead :

a) We’ve made patches, Calendly is a neat new way to arrange a meeting, but it won’t book taxis to the meeting than move if we cancel the meeting. It won’t fill in timesheets for us, it won’t book a meeting room for us. Quickbooks is a nice way to send and chase invoices, but it won’t “take care of invoicing” in the most holistic sense.

b) We’ve made little OS’s, like Asana or Notion, Trello or Basecamp, which act as a way to bundle some tasks into one interface, but these programs really act like dashboards on top of broken systems and simply provide reminders and management more than doing the work

c) We’ve made integrations and used API’s to allow programs to pull from each other, we’ve software like ClickUp or Monday which can process PDF’s, access files on Gmail, or Teams or OneDrive, it can in theory use leads from Salesforce to send emails in Mailchimp, to update calendars on Gmail.

d) And we’ve even now got Workflow Automation, like Zapier, which in theory can perform set tasks, in set orders, with integrations and take care of entire flows like “ get new inbound lead from Google Sheets, act contact to HubSpot, send them DM on Slack, send them an email if they don’t reply”

And this is reason numher one why AI won’t take our jobs. Our jobs, even the most routine, most basic, most “computery” are still a mess, involving judgement, emotion, copywriting, compassion, empathy. And they take a long time because the foundations on which we made our work, were wrong.

AI can indeed do “Action items” but it can never do Tasks, let alone do objectives, let alone do a job.

If AI is to take jobs, it will do so on this basis.

1. There are not many, but there are some entire “fields” of work, like photo retouching, or mock image generation or perhaps “writing code” or language translation , which are jobs where 80% of someones job maybe simply to do “one” task, and one thing that basic AI can very well. But most people who understand AI, don’t understand how rare these jobs are, a lawyers job is not to “write contracts”, an Architects job is not to “design a building” , these are typically tiny elements of a job which is really about understanding needs, finding new clients, solving problems creatively, training staff around you, etc.

2. Companies will entirely restructure around AI.

In theory, smart, efficient companies would undertake time and motion studies now and find out how work is done by the entire company. It would then go through the process of finding ALL of the tasks done by ANY single person, and find a way to group these less around roles and people, and more around how they are best done.?Rather than having 100 people writing occasional trends pieces, you may create a team of 3.

Rather than having 50 people managing the entire supplier process, you may make a team of 3 people who just chase involves all day long. 3 people who just on board new suppliers. If companies are to do this ( a rather soulless process) , they may find out that AI can be used to effectively used to deal with some of the tasks once extracted from a process like this.

And so long as the interfaces between people and technology make sense and are efficient. I’ve already had an invoice take far more work by ( more senior than normal) humans to process, because a company tried to use an automated image recognition based system.

3. Roles could be done in different ways.

In the image above, A French teacher is the most safe. The act of a person in a room giving lessons, looking after kids, writing reports, answering questions, is indeed not something a robot can do. But if we recast the role of teachers less about taking care of kids, and more about the simple transactional value. One could argue in the age of AI, a robot won't teach French, but French won't be taught.

But the reasons this won’t happen are far far more interesting.

1. Most companies are about power more than efficiency, and power typically comes from having big internal budgets and employing more staff. Both of which are more powerful than the case to be made for automation. Power is what drives company dynamics, not profit.?
2. Companies have a terrible track record of using even basic technology to improve how they do things, mostly because companies are made up of people and people are strange. I’ve known places where booking a meeting room took phone calls and ages, because it was just ‘how things were done around here”. Many people today still use paper calendars not shared digital ones. If companies care about AI, why didn’t they care about using Excel properly or training staff on how to use email better?
3. AI won’t be as good as people think it will be. So far it’s mainly a way to do things badly but fast and free, most processes in business are should ultimately be about better, not easier or cheaper. ?Yes, Zapier can automatically cancel a meeting, but what if it emails someone who is grieving a lost family member with the same tone as everyone else. Life is more complex than we make out, relationships matter too much to outsource feeling to machines.

Interesting times ahead. I feel personally most companies have more to gain by thinking more about what they shouldn’t do with AI, but just things they shouldn’t do at all

I’m more excited about companies with clear processes to follow, companies who run meetings well, companies who focus, companies who make decisions fast. It’s those who will thrive first.