There is a lot of excitement surrounding the new software made possible by generative ai. Applications like natural language chatbots that respond intelligently to complex questions with custom knowledge pique interest because these applications were previously impossible. Generative ai has very quickly enabled a whole new set of applications, and this is justifiably exciting.
Less exciting but equally impactful will be the replacement of traditional, custom-developed software with general purpose LLMs to solve software problems that we could already solve, but can now be implemented without problem-specific code. Generative ai is highly flexible and with a bit of prompt engineering a large language model (such as GPT-4o) can be used to solve a significant number of traditional software challenges. While it is exciting to build new things previously impossible (or highly infeasible), much of the impact of generative ai will be building differently what we can already build today.
This shift from problem-specific application development to integrating problem-agnostic, generalist generative ai will change dominant architecture patterns, organizational skill set requirements, and cost-of-development equations. It is the reason why every development…