The landscape of software engineering is on the brink of a significant shift, propelled by the advent of large language models (LLMs) deployed as generative AI. Notable players in this arena include ChatGPT and emerging alternatives like Google Bard. The integration of AI in software engineering is inevitable, and the refusal to adopt these advancements could lead to obsolescence. These AI tools can either enhance our problem-solving abilities or expedite the deployment of code.

Reflecting on the past, we can draw parallels with the introduction of Dreamweaver, a tool designed to automate website development. Despite its promise, the code it generated was often messy and required manual editing before deployment. However, it was faster than writing raw HTML and CSS from scratch. Today, web developers have a plethora of tools at their disposal, from Visual Studio Code to WordPress. Yet, the best results still require human intervention. These tools have not made human skills irrelevant; instead, they have augmented developers' abilities by automating repetitive tasks and catching common errors.

The rise of LLMs and generative AI marks a pivotal moment. Tools like GPT-3, GPT-4, ChatGPT, and Google Bard have transitioned from being fringe curiosities to subjects of cultural fascination. GitHub Copilot, built on the OpenAI Codex, is essentially a version of ChatGPT specialized in software engineering. It streamlines tedious tasks, offers intelligent autocomplete, detects patterns, and generates error-handling code. However, it's not infallible and often produces incorrect code, reinforcing the need for human oversight.

Generative AI will amplify what we prioritize in software engineering. For those who prioritize speed over quality, LLMs will increase the rate at which they can deploy subpar software. For those who view code as a commodity, LLMs will automate the assembly of derivative software, leading to stagnation. However, for those who value innovative solutions and critical thinking, LLMs offer a more satisfying potential. They can form a hybrid partnership with human engineers, combining their strengths to craft secure, stable, and sustainable solutions to important challenges.

We are in the early stages of this AI revolution. LLMs are raw materials with untapped potential. To harness this potential, we need to train these models to generate high-quality code and adopt test-driven development. We also need to learn how to maximize the benefits of our new AI collaborators.

In a decade, we envision a future where good engineers become exceptional hybrid engineers, working in tandem with AI to solve problems and create solutions more powerful than either could achieve alone. The key to this future lies in responsible training and application of AI. If done right, our AI collaborators will not replace us but will make us better.