Will AI-Assisted Software Engineering frustrate Innovation? Is there bias in Tech Choice
AI Assisted software engineering (with the likes of GitHub Copilot, Cursor, Windsurf, Claude, Roo Code and LLM of your choice) maybe bad for innovation, I fear. Why? Because these assistants are much better at providing support for mainstream technologies than for new initiatives or niche product.
Large, mature technologies for which many resources are available for training — code as well as documentation — are presumably much better placed to get huge acceleration from AI assistant, then niche languages, frameworks and libraries. When a developer is looking for a high productivity boost from the AI assistant and the AI assistant is best capable of providing that boost for already established technologies, what is the incentive for this developer to try out a new, fledgling technology? A freshly minted framework? A beta of a small scale open source project?
I am afraid that we may be in danger of losing a lot of evolutionary potential , of stifling innovation. Of smothering new initiatives . Because our AI assistants are not innovative themselves.
Am I too gloomy about this? Or at least can we do something about it? By doing research ourselves and guiding (prompting!) our assistants to new tech we like or at least want to give a chance? By exploring the new technologies even more than we already did — and leveraging the AI tools to do such research for quickly and efficiently? By aiding starting technologies by creating documentation, example code, articles — even using AI tools — to provide more material for the AI assistants to learn from?
AI-assisted software engineering holds immense promise and it feels almost like magic to conjure up new applications using a tool such as Windsurf Cascade. But it also comes with challenges that we must address thoughtfully. By proactively exploring new technologies, contributing to their development, and maintaining a balanced approach, we can harness the benefits of AI while ensuring that innovation continues to thrive in our industry. We need some form of DEI in recruiting technologies to prevent always ending up with the white male ;-)
Am I too somber about this? Perhaps. I hope so. I am curious to learn how you may feel about this. Especially if you are involved with an open source project or a niche framework, library or tool that you feel is threatened by mainstream oriented AI tools.
