Become a 10x developer by harnessing the power of Generative AI

Will Generative AI replace developers?
No, it won’t, but developers who know how to use it in their daily work will likely replace developers who don’t.

How could you leverage Generative AI tools, like ChatGPT and GitHub Copilot, to write better software faster?
This session presents how to use Generative AI tools to transform software development, highlighting their strengths and weaknesses.
We will give some demos using a coding copilot to write Java applications. After this session, you will be able to delegate all the tedious development tasks to Generative AI so that you can focus on all the exciting stuff in software development!

This recording is from the presentation delivered by Ioannis Kolaxis, Director at Accenture, at the JPrime conference on 28 May 2024 in Sofia, Bulgaria.

0:00 Electric refrigerators disrupted ice distribution. Will AI coding assistants replace developers?
1:44 What will you learn from this presentation?
2:18 Who am I? Short Bio
2:56 Do you use AI coding assistants?
3:14 Weaknesses of AI-assisted coding
4:09 An example of a coding prompt
5:35 Demo: GitHub Copilot in action, generating a Java method based on the code prompt
6:09 First weakness of AI coding: Generating new code that is functionally incorrect
7:45 Which coding prompts are more likely to generate functionally correct code?
8:37 10x developers break big, complex problems into smaller, simpler ones
9:45 Second weakness of AI coding: Refactoring will likely generate incorrect code
10:57 Common failures in AI-assisted code refactoring
11:30 10x developers verify AI-refactored code is correct by running unit tests and reviewing code
12:10 Where do we spend most of our time as software developers? Writing new code or reading existing, legacy code?
12:57 We spend 70% of our time reading code and only 5% writing code
13:39 Third weakness of AI coding: It makes it harder to maintain code due to an increase in duplicate code
15:57 AI coding assistants do not promote code reuse
18:22 10x developers prefer reusing code than copying/pasting code via AI assistants
18:44 Strengths of AI-assisted coding
19:14 First strength of AI coding: Helping explore new technologies and develop prototypes
21:53 10x developers use AI coding assistants to move fast and break things
22:28 Second strength of AI coding: Understanding code and enhancing its performance
23:13 Demo: GitHub Copilot identifies sorting algorithm, and recommends alternative more efficient algorithms
25:18 Refactoring code to use efficient algorithms can drastically improve performance
26:02 10x developers use AI assistants to understand code and identify opportunities to improve its performance
26:20 Third strength of AI coding: Generating unit tests that act as a safety net before modifying existing code
26:40 Demo: GitHub Copilot generates unit tests for a Java method using the JUnit framework
29:08 10x developers move fast and don't break things by using AI coding assistants to generate unit tests
29:23 What is the future of software development?
29:41 Demo: ChatGPT describes a photo of a dog
30:34 Large Language Models (LLMs) enable us to build software that was previously impossible!
31:58 LLMs vs. Java applications
34:00 Coding Challenge: Build a software application that finds the cheapest flight tickets to travel to the upcoming Oracle JavaOne conference
35:11 Demo: A custom ChatGPT from Kayak that finds the cheapest flight tickets
37:03 Codeless: A new way of building software applications
38:13 Summarizing what we learned for AI-assisted coding
39:29 Summarizing software development paradigms
40:19 Will AI replace developers?
40:28 Airplane pilots: the humans in the loop who focus on the broader aspects of operations
41:08 Software developers are the humans in the loop, focusing on the bigger picture of software applications
42:50 How to harness Gen AI to become a 10x developer
43:48 Questions & Answers

#AICoding #AISoftwareDevelopmentTools #GitHubCopilot #ChatGPT #genai #generativeai #codingassistant #codingassistance