Digital Wingmen: The Evolution and Impact of AI Co-Pilots
The emergence of ChatGPT and its rising popularity has fuelled the dreams of building an AI-enabled digital assistant that automates (read: partly automates) all mundane tasks that an individual has to complete over a finite period of time. “Co-pilot” is the term that has become the accepted lingo for such digital assistants.
Fundamentally, a co-pilot is someone who assists a captain in performing tasks required to fly an aircraft. Typically, the tasks and qualifications of a captain and co-pilot are the same. Captains decide the tasks they should do and the ones co-pilots should do. The need for a co-pilot arises when the captain’s ability to perform well gets compromised because of some constraints including time, complexity, or standards. Given the risks associated and outputs expected with the job, having a co-pilot onboard becomes a preference.
In the digital world, “standards” are driving the need for an AI co-pilot. The want for a quicker turn-around to deploy a product in the market, and higher engagement through rapid content creation has persisted for almost a decade. But AI now seems to be the messiah that will help us reach this utopia. And what is this utopia you ask? For most, it’s an “auto-pilot” — an assistant that does everything for you without you lifting a finger. In tech terms, it is artificial general intelligence — an AI that has the capability to improve itself. Although we’re quite not there yet, GPT4-enabled co-pilots have put us a step closer to achieving OpenAI’s long-term vision.
Now the question is — are these standards realistic? Does every task and action require a co-pilot workflow? In my previous writeup titled, “The AI Impact Framework”, I introduced a framework to map the areas where AI is actually useful. The 4P framework gives a direction to evaluate AI applications based on 4 parameters — people, productivity, programmability, and precision. Tasks involving a large number of people, spending a lot of time doing the task (productivity), with a digitized workflow (programmability) and limited scope of error (precision), are the ones that would be transformed using AI. A co-pilot UX is ideal for high-precision tasks where users have a digitized workflow. The margin for error of such tasks would be low. An autopilot would be ideal in digitized tasks where the margin for error is high and human intervention is not required.
Ideal user experiences to interact with AI
Based on the framework, we can assign the ideal AI UX to different use cases as per the required programmability and precision. Currently, due to the unavailability of auto-pilot technologies, most of the use cases for which auto-pilot is the ideal AI UX are being built with a co-pilot UX.
Ideal user experience to interact with AI for different use cases
Current AI Copilot Ecosystem
Current forms of co-pilots are primarily Chrome extensions, with some exceptions of plugins for code generation. Trending co-pilot projects can be classified into 3 categories — Co-pilots for software development, business & enterprise use, and general-purpose co-pilots.
Applications in Software Development
AI is being employed in software development use cases where generative models are used to generate new lines of code or complete existing incomplete code snippets by entering a prompt. Different co-pilots have gone one step ahead to translate code from one language to another and debug errors in code.
Applications in Enterprises and General Purpose Copilots
Different functions in organizations are adopting AI to improve productivity in their respective domains. Microsoft recently announced a copilot for its Office Suite. IT professionals are managing identity and privileges granted to employees using AI copilots.
Organizations like Adept and Bing have created general-purpose AIs that can be used by anyone to automate their digital processes.
As we navigate the vast, unknown expanse of AI’s future, the concept of a digital “co-pilot” emerges as a promising and tangible reality. Currently limited to specific tasks and domains, these co-pilots are nonetheless transforming our approach to work and productivity, freeing us to focus on more value-added tasks. While the transition from co-pilot to auto-pilot technologies presents significant challenges, the expanding co-pilot ecosystem signals a step closer to the utopia we envision, a world where AI handles our mundane tasks, and we engage in creative and critical thinking. As we continue to refine our AI Impact Framework and responsibly innovate, we move steadily closer to the horizon of AI applications, anticipating the evolution of today’s co-pilots into tomorrow’s auto-pilots.
