Leveraging A.I.

A.I. is obviously top of mind for many companies and investors as it is an emerging technology that is demonstrating its purpose and usefulness to a degree that no other emerging technology has in the past. Many forecast this to take the jobs of developers and other professional services however we have a different perspective on this. We believe it is an absolutely amazing technology that must be embraced, promoted and pushed forward however it is has a long ways to go to start taking development jobs, in fact one could argue that it never could take over these roles at their higher level positions.

Why could it never take over these roles? Well for 2 primary reasons, first being the amount of compute power it would take for A.I to fully assess full projects/systems in order to make the proper decisions; it has only proven itself to really be useful with small apps. Secondarily, and the most important reason is that A.I. is nothing more than a replication of the collective historical decisions which means it would never really effectively lead new technologies and innovations because it simply could never understand it without enough historical information to direct it.

We attempted to whole heartedly disprove the statements above and tried to create a series of applications by only directing A.I on what to build, starting at the application structure, all the way down to each page/element. What we found out was the even with historical data, A.I. struggles with dependencies and their relative structures, how/when to use the appropriate libraries/code and how to efficiently code at a higher level. We spent more time trying to understand why/what it was doing and then course correcting it so our projects would do the simplest of things such as compile.

However, the one thing we did discover is that it is an amazing tool to be a force multiplier during the development process, and more-so during the prototyping phases of our projects where the actual implementations are less important and the focus on a partially functioning proto-type is the goal before the real development starts. We have collectively come to the consensus that when using A.I. to assist with development, refactoring of small sections of code, building unit tests and generation of new micro boiler-plated content, that we could easily increase our velocity from 20-40% based on the type of projects we are working on.

We have whole heartedly adopted A.I assisted programming in almost every aspect of what we do now so that we can take the burden of writing simple sections of code off of our shoulders which keeps our velocity higher than we normally operate at with the least amount of effort.