What AI means for software development

Artificial intelligence (AI) and what it means for jobs, industries and sectors has been a hot topic in the media, Westminster and in offices up and down the country for some years.

Many fear it, seeing only the possible impact on jobs; others see potential for speeding up and streamlining processes. For software developers, it’s not a major threat, in fact it offers new opportunities, but those opportunities bring with them insurance risks that we’ll help you understand.

Getting there faster

With just a prompt, AI tools like Chat GPT, GitHub Copilot and Amazon Code Whisperer can generate software code, but these aren’t complete solutions and won’t replace software developers.

AI-generated code needs human review and debugging, it lacks deep understanding of business logic or system architecture and it may produce insecure, inefficient or outdated code.

What it can do is help with boilerplate coding where sections of code are standard and repetitive. Here it can help developers save time, meaning they have more time for the complex logic or architecture.

Deploying AI for testing

Traditionally, testing is performed manually, by analysts. Testing software for a variety of scenarios can be time consuming and is subject to human error. Here there’s an opportunity for AI to make a real difference.

AI tools can help identify bugs and fix them faster. However, developers need to beware. AI tools are continuously self-learning, using your data and code to train their system. Sharing your data and code always comes with a risk and that’s no different when you’re sharing it with an AI tool.

Widening access

Tools like no-code or low-code platforms offer exciting possibilities for widening access to software creation.

Enhanced by AI, these platforms allow non-developers to build apps or workflows and can help developers rapidly protype ideas.

Widening access means faster innovation too.

What are the limitations and risks?

AI-generated code still needs a person to review it and to date, AI continues to struggle with understanding the context and business logic of what it’s doing.

When it comes to risks,there are two significant issues yet to be overcome:

Security is paramount and there remains a risk that AI-generated code could result in insecure code patterns. Hardcoded credentials, SQL injection vulnerabilities and improper input validation are all possibilities with AI-generated code and using it without a review process can lead to exploitable flaws.
Using AI to generate code brings with it uncertainties over ownership and licensing. Without careful checks, you could open yourself up to allegations that your AI-generated code has been copied or stolen from a third party and the result could be disastrous for you and your business.

The bottom line

It’s not likely that AI could or would replace software developers any time soon, but it can be useful, particularly to engineers who want to save time in development and testing and who understand the risks and can implement safeguards.

Contact our expert team if you want help understanding the risks associated with using AI.

What you need to know about insurance for software companies