Creating Super Humans: How the marriage of technology and human intellect can help your business grow
We’ve all read the alarmist headlines: Automation will displace 38% of U.S. jobs by 2030. If true, the implications are hard to grasp. Would over a third of the population be out of work?
But then there’s a counter narrative: we can find ways to use advanced technology to amplify human intelligence. In this counter-scenario, the jobs wouldn’t disappear, but rather, each person’s job would use a combination of human and technology, creating a higher level of output. In this way, we would create Super Humans.
We’ll touch on the implications for the future of work later, but first, let’s discuss how to use a framework to decide which parts of your business processes are ripe for technology now, which ones are best for humans, and which ones should use both.
We’ve benefitted from this framework, at Radicle, a disruption research firm (yes, a startup that analyzes startups :) that I joined within weeks of its founding and for which I have had the fun of designing processes and systems that empower our team of analysts, developers and data scientists. Here’s what we learned by developing a system that produces high-quality research through the marriage of humans and technology.
Aside: Terms like “artificial intelligence” and “technology” can encompass a spectrum of different meanings. For our purposes we use “technology” to describe a range of computer science tools or components that one can employ to solve different business challenges.
Creating Super Humans: How we did it
We started with a product that customers loved and were willing to pay for. Then, we worked backward, dissecting it into groupings of microtasks. Finally, we applied classic and novel process design principles to separate the pieces ripe for technology from those that were more efficiently done by humans. At the end, our re-assembled product seamlessly blended technology and human intelligence, giving our team members “Super Human” capabilities.
The production of research is a surprisingly complicated process that requires balancing interactions between various teams and managing multiple input resources (different analysts, external and internal FTEs, etc.). However, these principles can apply to anything you make: software, hardware, physical products, or any services.
The framework has four pillars:
1. Balance machine-work with team-work
2. Automate the mundane
3. Tackle the hard
4. Create a virtuous cycle of improvement
1. Balance machine-work with team-work
We started with the common thesis that technological tools would enable our company to scale. However, for our research product, the primary challenge we faced is that a fully human approach leads to inefficient outcomes, while a fully-automated approach results in research without nuance, just scores and other algorithm-driven mumbo jumbo.
We had to develop an approach in which humans and technology could together produce a product that was more valuable than either could produce individually.
To identify how best to combine technology and human intelligence we split our product into a hierarchy (job > tasks > microtasks).
Once we had our list of microtasks, we thought creatively about where we could use technology (the tools that we developed/deployed were: automation, coding scripts, natural language processing & machine learning). Then, we compared that to the status quo of having a human perform the activity. We asked ourselves two questions:
- “Can we use technology to complete this task?”
- “What would technology entail from a cost/benefit standpoint (time, money, or opportunity cost)?”
Can technology do the task better and more efficiently? If so: automate!
2. Automate the mundane
Dividing tasks into skilled and mundane gave us a framework to split tasks between humans and machines. It also gave us an added people-pleasing bonus. Mundane tasks (repetitive, recurring, straightforward tasks) are best to automate because:
- It’s cheap and easy (with today’s tools)
- Humans hate the mundane
Mundane tasks, by definition, follow a simple flow: do this, then do that, then do this other thing, repeat. Once you know the formula, there is an abundance of software to help you write the equation for the task and execute it (e.g.: Excel’s/Google’s macros, Zapier, IFTTT and even simple programming). For these, the “hard” part is writing the formula, the tools make it simple to implement with no or little technical expertise.
The higher-level benefit comes from a human resources perspective. Modern employees are well trained; they’ve been taught since high school to think analytically. Simple repetitive tasks, therefore, become boring and feel like a waste of time. By automating the mundane tasks, your company is providing a higher purpose: employees feel liberated to pursue more “value-added” jobs (better work) or even have a bit more free time (more leisure).
Automating the mundane also helps with the internal perception of your company and the feeling of overall progress. “I used to have to do X, which was a pain, and now I don’t have to — we’ve gotten better!” Good employees will recognize that you invested in them, and reciprocate — improving your company (and the bottom line)!
3. Tackle the hard
Through the process mapping, we quickly found that not all tasks are mundane, many are very, very hard. For example, the task of discovering new companies and precisely grouping them into logical categories (industries, sectors) is very difficult. So, what to do? Should we use only humans? Only machines? Or both?
Our approach: we paired a human with a machine to find what we needed. First, a human identifies a couple of prototypical companies. Then, we fed information on those prototypical companies into a sophisticated machine learning (ML) algorithm. The MLA “crawls” through a giant data set (with 80,000 companies) and surfaces a small list of relevant, related companies.
The result: a complete, comprehensive and thoughtful list of similar companies (and in turn, a strong foundation for further analysis). And importantly, a system that is better than a machine or human could achieve individually.
4. Create a virtuous cycle of improvement
Once you have a process set up, you’re done, right? Time to send the hackers home. Not quite yet…
Using this process will set your company up for operational success, creating a great foundation and freeing up time to tackle the next set of challenges. However, things change, and any good system needs to be adaptable. To do this, we employ 2 tactics:
- Adopt continuous deployment for process improvement. Empower everyone involved in the system to learn and implement design principles at each microstep. Encourage them to add efficiency steps at the atomic level for which they are responsible / have good visibility. More heads, hands, and hearts are better than one. Set regular check-ins, watch for company growth and system strain. The system will steadily improve each day, and exponentially improve as you grow.
- Keep on the lookout for “helpers”. Products are released daily to help businesses. Technology improves exponentially. With your first end-to-end process defined, you know precisely what you need (thanks, process mapping!), and you can immediately see if a solution will help you or not. And with an existing process in place, you can easily evaluate if new tools are accretive because you have a point of reference.
Thoughts on the Future of Work
The future of work will see humans increasingly working alongside machines. While the long term effects are unclear, historically, humans’ ability to “catch up” to technology is immense.
Having a framework to understand how to leverage the talents of both humans and machines optimally will enable you, and your company, to realize solutions that are better than the sum of the parts. But to achieve balance — or dare I say, harmony — between technology and humans, we have to approach process design with the view that the most elegant systems may not be all human or all machine but, instead, some of both.
Image Credits: Robot & Frank / Kate Venezia
