Copilot X: AI Pair-Programmer is just out. What's next?

Last week GitHub announced its Copilot X,?AI pair-programmer for developers. It offers auto-complete-style suggestions as you code. It's practically a code-centric chatbot, powered by OpenAI's GPT4 that helps developers write, test, and debug their code.

Studying the genesis of modern software development platforms, I conclude that Copilot X is a combination-innovation of the following concepts, that are evolving over the past two or more decades.

1. Code Sharing: In 1998, Netscape announced that they make the next-gen communicator source code available free to harness creative power of thousands of internet developers. This not only paved ways to seed a global net-centric market but also accelerated the thinking around code sharing, open source etc. Copilot enables the developers with access to sample code, improving their productivity many fold.
2. IDE (Integrated Development Environment): In early 1990s, Turbo Pascal was spoken as the "next big thing" for developers. It had the idea of an integrated development environment. It provided a source code editor, all the tools necessary to compile, run and debug the code, and to build the executables, so the developers need not 'run-around'. However, Microsoft Visual Basic was the first one to live up to the definition of an IDE. Copilot is a sophisticated IDE that provides many weapons that developers need today.
3. Network Effect: Network effect is a phenomenon whereby a product or service gains additional value as more people use it. You need a platform to create the network effect. A platform is the ecosystem that hosts the producers and consumers for interaction. The concept of a platform is centuries old. GitHub, for e.g., is an online software development platform, that developers use to share code and work on projects together. Copilot with the network effect of 100 million developers can contribute to many more disruptive innovations in the near future.
4. Artificial Intelligence: AI is a topic that the human kind has been researching for over a century. It has now become mainstream. Generative AI which is a type of AI capable of generating text, images, or other media in response to prompts. Copilot generates code to developers queries so that they spend time on more value adding activities, than working on mundane pieces of code.

Without assistive technologies, quality of life or business cannot be improved. Resisting to change or ignoring the values of the innovations will only lead to 'Kodak moment'.?Generative AI presents such an occasion where we get excited with the value of the innovation and at the same time, concerned with the variety of its challenges such as performance, security, IP, warranty, operational inefficiency, poor dev practices, legal/ethical challenges due to plagiarism etc.

Three specific areas that we should fast get a control over are (1) Data Management/Governance and (2) Testing and (3) Talent & Culture.

Data Management and Governance

Let us get the facts straight - Code may be autogenerated whereas (production) data is not.?While code is the body, data is oxygen. Microsoft and EDM council have done some meaningful work around data management capabilities that are critical as we accelerate the migration of data to cloud and fast adoption of AI. It is important to provide comprehensive data management automations and controls for protecting sensitive data to accelerate trusted cloud and/or AI adoption. Microsoft Purview, Informatica data quality, Oracle Enterprise Manager, etc. are examples of solutions that enterprises should start adopting to put the standards and controls in place.?

Testing?

Traditional testing methodologies and tools will only compromise the values that you expect from the CoPilot adoption. When you start to use generative AI to produce code, you need AI to produce the test scripts as well. Codeless programming needs codeless test automation. AI Pair-Programmer needs AI Pair-Tester. You need AI to test AI.

Ecosystem providers such as?ACCELQ,?LamdaTest,?Sauce Labs,?Tricentis?etc. are all having low-code/no-code testing platforms that simplify test creation, orchestration, and scalable test execution for easier collaboration among the stakeholders for faster, higher-quality, and more durable releases of applications to production.?They help you adopt continuous testing, in your existing software release toolkits and provides a fast, reliable, secure, and scalable way to manage end-to-end test execution so that you can seamlessly run tests from Playwright, Azure DevOps, GitHub etc.

Talent and Culture

McKinsey in a study, reports that "When an organization manages both its talent and culture effectively, the interplay between them can create a virtuous cycle: attracting talent, sparking innovation, retaining talent and creating impact. Culture is the most significant self-reported barrier to digital effectiveness. To find, hire, and retain these employees, companies must build their technology organizations around the right leaders, explore new ways of hiring, create career paths that fit technology talent, and transform their culture to facilitate the work". This is the best answer to the question of what organizations should do to adopt generative AI in terms of its talent strategy.

The market size of Generative AI is mind-blowing. If we want to avoid our companies going through a 'Kodak moment', we better know how and when to abandon traditional business and technology practices and take advantage of the new waves of innovation.