I’ve attended CITCON once before, but 2024 was the first time I took to the stage, and it did not disappoint. The open-space conference was as lively and enthusiastic as ever and it may not come as a surprise that the main focus was AI.
Before the COVID-19 pandemic, AI was mostly defined by machine learning and data-driven statistics. However, roughly two years ago, large language models (LLMs) took the world by storm, profoundly transforming the AI landscape and our everyday lives.
Is AI a blessing or a curse?
The user-friendly delivery of LLMs has reshaped how we do software development. At this year's conference, it was clear that AI was at the forefront of everyone's mind, including myself.
That's why I proposed the topic, "AI: A blessing or a curse?" My aim was to spark discussions on how AI tools like ChatGPT and GitHub Copilot are changing our work and to explore both the benefits and challenges they bring to the table.
The current usage of AI
The session drew in a dozen or so attendees, and I opened with the question: “Have you used AI-related tools, and if not, why?” This allowed everyone to share their experiences. Interestingly, only one person mentioned they hadn’t used ChatGPT or similar tools before, and this was down to data and security concerns. The rest had various uses for AI.
Most participants used it at work and in their personal lives. For example, some used ChatGPT to generate food recipes, appreciating the convenience of bypassing lengthy articles, books, and website ads.
Personally, I use ChatGPT for my Dungeons and Dragons homebrew campaign. While I know the major plotlines, filling in side quests and other details is much easier with ChatGPT. Many others mentioned using AI for overcoming writer's block, business communication, document generation, and code and test generation.
An interesting use case was finding the proper words based on domain language. For example, some participants used AI to come up with meaningful upper-level names or suitable ones for automated test cases.
The benefits of using AI
After sharing personal use cases, we discussed the benefits that participants experience using AI tools, such as:
- Improved efficiency: AI provides quick answers and reduces the time needed for deeper research or modifying existing answers to suit specific use cases.
- Enhanced quality: AI offers suggestions to improve readability or usability, resulting in higher quality outcomes.
- Task simplification: AI simplifies tasks by creating easy-to-write code for developers, allowing them to focus on more complex aspects of their projects.
- Creating new jobs: AI has led to the creation of new roles, such as prompt engineering, demonstrating its impact on the job market.
Downsides and worries of using AI
“With great power comes great responsibility,” and AI tools are no exception. Unfortunately, these tools don’t always present a win-win situation, so we also discussed the negative impacts we’re seeing when using them.
One troublesome issue is that AI tools can “hallucinate,” producing incorrect or even outright false information. Escaping this rabbit hole isn't always easy. Another critical concern is how AI models are trained and whether the learning data is always coherent. We also questioned when we might reach a point where learning data is not provided by humans, leading to an AI-to-AI learning loop.
Personally, I've noticed a side effect in the form of tiredness. I treat the output provided as if it were written by someone else on the internet, requiring careful examination. This tiredness also seems to stem from increased efficiency. Previously, writer’s block gave me a chance to think about the problem before seeking more information. Now, I can ask a vague question and start iterating from the AI's response or redo the prompt.
Other participants raised concerns about the future careers of junior or less experienced developers. Although current AI tools have created new roles, they also reduce the need for simpler development tasks that AI can handle much faster.
Key takeaways from my talk on AI at CITCON 2024
The future role of education was also brought up. Is it necessary to learn multiple programming languages in school, or should schools focus more on the basic principles of programming and the general management and maintenance of software projects and architecture? Is the cognitive load of constantly reading other people's code and text something that should be raised more as a topic?
Some participants noted that companies have started to open-source their closed-source projects to benefit from AI tools that use open-source repositories as learning data. Personally, I find this trend both interesting and concerning. Could this shift mean that brand value becomes more important than intellectual property rights?
Most participants agreed that some form of governance to control and facilitate AI development is necessary. The EU AI Act, adopted in March 2024, provides guidelines on using LLMs and other AI tools and aims to create a regulatory framework.
Circling back to the initial question: Is AI a blessing or a curse? The consensus from my talk was that the benefits currently outweigh the drawbacks, using the analogy of fire: It can be a great tool but a poor master.
Published: Jun 11, 2024
