What types of work CAN be assigned to intelligent machines?
Part 3 of a 7-Part Series on Using Collaborative Intelligence (CI) to Improve Company Performance
As I wrote previously, when speaking with company executives regarding how best to use Collaborative Intelligence (CI) to improve their company’s performance, I ask seven clarifying questions, the third of which is: “Many types of work can be assigned to Intelligent Machines (IMs). What type of work do you hope to assign to them?”
I ask this question to describe six characteristics of task-level work and help executives think more clearly and thoroughly about where and how their company can deploy IMs most effectively. Thinking more deeply about the work is essential because the nature of the work determines whether it is a good idea to assign such work to an IM.
The table below lists the six characteristics and their corresponding attributes. The list following the table provides additional detail.
Keep the following points in mind when reviewing the table and list:
Completing a specific task may involve multiple attributes of each characteristic, but the focus should be on the predominant attribute. For example, in the case of the Labor characteristic, if the task is Write a response to a customer inquiry. the crafting of the response is Mental, and the typing of the response is Physical, but the task would be characterized as Mental.
There are more fine-grained attribute lists for each characteristic, but adding additional options makes analyses much more time-consuming without a commensurate increase in value/insight.
A standard comment regarding the applicability of IMs to each characteristic is, "As the rate of innovation in this area increases, the scope of applications will increase accordingly."
Subsequent articles will consider whether tasks currently completed by humans should be assigned to IMs by considering the costs, benefits, and risks associated with such assignments.
Task Characteristics/Attributes
Labor: The type of labor required to complete a task is Physical or Mental.
Robots can complete or augment specific physical tasks completed currently by humans. AI and other forms of machine intelligence can complete or augment specific mental tasks.Work: The type of work required to complete a task is either Procedural (done exactly or much the same way each time) or Variable (there may be a common approach across tasks, but task completion requirements vary in each instance).
Automating procedural/routine tasks is a well-established practice. Robotic Process Automation (RPA) is one technology extending such automation's scope. Large Language Models (LLMs), such as ChatGPT, have revealed an ability to complete or support the completion of variable tasks.Output: Outputs created via task completion can be Standardized (such as those produced by manufacturing processes), Structured (variability within a specific form, ranging from a loan application to a customized product/service), or Unique (outputs may share characteristics but are essentially one-of-a-kind due to variations introduced in the creative process).
IMs excel at producing standardized and structured outputs, especially when the processes that yield structured outputs follow specific rules or standard processes. IMs, especially LLMs, have demonstrated some forms of creativity, but the types of outputs are constrained by current capability and other limits.Style: The working style required to complete the work is either Independent (the work can be completed independently of other work or other people, such as digging a ditch, welding a quarter panel onto a car, or composing an email) or Interactive (the work requires more than one person to participate in the completion of the task, such as crafting a new branding strategy or designing a new home).
IMs currently work independently of and interactively with human collaborators.Location: The nature of the work requires those engaged in it to be Co-Located (such as when providing personal services like a haircut) or Virtual (participants can be anywhere, such as when conducting a meeting or co-authoring a document). Please note that even if an IM replaces a barber or manicurist, the person receiving the haircut or manicure still needs to be co-located with the IM.
IMs currently work in co-located and virtual environments.Timing: The timing required to complete a task is either Synchronous (must be done simultaneously, such as when performing orchestral music) or Asynchronous (can be done at any time and, perhaps, in any order, such as writing a book or creating a software product).
IMs can be deployed to support the completion of synchronous and asynchronous work.
Based on its specific nature, all work falls somewhere along the following continuum:
Already: IMs can be and have already been deployed to complete this work. Examples include Credit Scoring and Algorithmic Trading.
Now: Due to recent innovations, IMs can be deployed to complete this work. Examples include computer code generation and content generation.
Near Future: Based on expected innovation rates, IMs will be able to take responsibility for some or all aspects of work currently performed by humans, especially in areas that require the combination of basic manual dexterity (repetitive tasks) and vision. Examples include cooking (basic recipes), mixing cocktails, and content generation using specific software packages, such as Microsoft PowerPoint® or Salesforce®.
Distant Future: Based on expected innovation rates and the nature of the work (and its complexity), it will take many years until IMs can complete such work well enough to supplant humans. I may be wrong, but I think the unconstrained use of autonomous vehicles (such that they can travel on any thoroughfare to complete any trip) fits this category.
Never?: The rise of LLMs and their apparent creativity has spurred an ongoing debate about the limitations of Machine Intelligence. Many argue that, while IMs are very good at applying knowledge, they will never be able to create new knowledge. For example, when will humans be able to ask an IM to create a never-before-seen framework for company executives to understand key drivers of business performance, such as Michael Porter created when he invented his Value Chain construct? I don't know what limitations, if any, will exist far into the future.