09-07, 10:15–10:50 (Europe/Lisbon), Auditorium
Discover the right helper for your Python coding needs. Explore security, code readability, cut-off limitations, creativity, ability to generate tests, and troubleshooting capabilities. Find survey results for programmers.
Given the recent boom of AI code assistants, more and more programmers are starting to use them to increase their day-to-day performance and delegate some routine tasks. But how to choose a suitable and robust helper? Is this possible at the moment?
During this talk, we are going to compare 3 AI code assistants for Python: ChatGPT, GitHub Copilot, and Amazon CodeWhisperer.
The criteria will be as follows:
- security: how passwords & SQL queries are being handled
- code readability: is generated code easy to follow?
- cut-off date: what is the latest knowledge of assistant (Python version, packages awareness)
- creativity: given a certain code, how good is the assistant at creating a Streamlit app?
- code testing: the ability to generate unit tests to code for different scenarios
- debugging/troubleshooting: the ability to find an error or to handle it
In addition to comparison, the results of the programmers' AI code assistants usage survey will be shared. By the end of this talk, attendants will have a good understanding of the differences between AI code assistants for Python, and they will be aware of the possible advantages/disadvantages of each mentioned assistant.
Data Scientist, passionate about data and its usage for problem-solving.
Darya's area of interest includes classical ML and NLP, LLMs, and GenerativeAI as well as working with AWS services. She is an active AWS Community Builder since August 2022.
Live-longer learner, taking part in online schools, courses, and workshops. Besides, Darya likes to challenge herself in hackathons.