ANZ Probes Disparities in AI-Assisted Programming Benefits
We independently review everything we recommend. When you buy through our links, we may earn a commission which is paid directly to our Australia-based writers, editors, and support staff. Thank you for your support!
Quick Read
- ANZ Banking Group is experiencing inconsistent advantages from AI-assisted pair programming.
- The bank is offering targeted training to improve familiarity with GitHub Copilot.
- AI-driven code generation is producing more recommendations than engineers are adopting.
- Certain specialized teams are not experiencing substantial benefits from incorporating AI.
- Despite initial suspicions, the overall quality of the code has improved.
Artificial Intelligence in Software Engineering: Disparate Advantages at ANZ
The ANZ Banking Group embraced GitHub Copilot early on, utilizing the AI-driven coding assistant to support pair programming. After a successful initial trial, access was increased from 150 engineers to 3000. However, the advantages have not been consistently experienced by all team members.
Uneven Benefits Distribution
At the recent GitHub Galaxy24 event, Chief Technology Officer Tim Hogarth discussed the matter. Hogarth mentioned that although some engineers have rapidly embraced and gained advantages from GitHub Copilot, others are still acclimating to the tool. “We initiated the rollout last year, making it accessible right away to 3000 individuals,” he stated. “We’re promoting onboarding and tracking usage activity.”
Developing Skills to Generate Value
To tackle this gap, ANZ is providing specialized training for GitHub Copilot. “During the proof-of-concept, participants showed enthusiasm for the trial, but not everyone is prepared to adapt,” stated Hogarth. Extra coaching sessions are being offered to assist engineers in forming new habits and seamlessly incorporating AI into their routines.
Specialized Groups and Restricted Worth
Despite these attempts, certain specialized teams have not experienced substantial benefits from utilizing GitHub Copilot. “We need to investigate this further,” Hogarth remarked. “We are uncertain whether it’s due to the technology, the team, or the nature of the problem they aim to address.”
Suggestions for Code: More Possible Solutions Than Approved Ones
Remarkably, GitHub Copilot has produced approximately two and a half times more coding recommendations than what ANZ’s engineers have actually accepted. This challenges the idea that engineers would uncritically adopt solutions generated by AI. “There was a certain ‘mystique’ surrounding Generative AI, where people thought you could simply input your problem and immediately get the perfect answer,” Hogarth elaborated.
Improved Code Quality
Contrary to initial beliefs, the quality of code produced by ANZ has indeed enhanced with the adoption of GitHub Copilot. “We’re discovering that it’s actually superior,” Hogarth affirmed.
Summary
Although the ANZ Banking Group has adopted AI-driven coding tools like GitHub Copilot, the advantages have not been uniformly experienced across its engineering teams. Specialized training and coaching are being offered to assist more engineers in effectively incorporating AI into their processes. Certain specialized teams are still finding it challenging to realize the benefits, and the bank is actively investigating the reasons. In general, despite producing more recommendations than implemented solutions, AI has enhanced the quality of code output at ANZ.
Q: How does ANZ Banking Group implement AI-assisted pair programming?
A:
ANZ Banking Group integrated GitHub Copilot for AI pair programming and increased its deployment from an initial test group of 150 engineers to a total of 3000 engineers.
Q: What causes the advantages of GitHub Copilot to be distributed unevenly at ANZ?
A:
The advantages are not equally distributed since some engineers have greater familiarity with the tool than others. To enhance understanding and utilization, training and coaching sessions are being offered.
Q: In what ways is ANZ tackling the issue of the unequal distribution of benefits?
A:
The bank is providing specialized training and extra coaching sessions to assist engineers in more effectively incorporating GitHub Copilot into their workflows.
Q: Is every team at ANZ gaining the same advantages from using GitHub Copilot?
A:
No, certain specialized teams have not experienced substantial benefits from utilizing the tool. The bank is exploring whether this is attributable to technological challenges, team dynamics, or the specific nature of their issues.
Q: What is the difference between the number of code suggestions generated by GitHub Copilot and the number of suggestions actually accepted?
A:
GitHub Copilot produces approximately 2.5 times more code recommendations than the number that engineers at ANZ actually approve.
Has the quality of code produced at ANZ declined as a result of using AI-powered coding?
A:
The initial concerns that code quality might decline have proven to be untrue. On the contrary, the quality has actually enhanced with the implementation of GitHub Copilot.
