Forrester has branded AI and generative AI for software development as TuringBots since its inception in 2020. The good news is that generative AI (genAI) has advanced the maturity of TuringBots by at least five to 10 years. So how has genAI changed the state of TuringBots today in the market?
TuringBots impact all companies in all industries, although the adoption pace across different industries varies. Technology executives in all industries need to learn how to deal with TuringBots. In fact, many experiments are run to learn quickly about the minimum viable governance and good practices necessary to avoid risk and compliance issues with unsecure or poor-quality code, slow performance, or poor user experience issues. This is a must, since developers will use TuringBots whether you like it or not — just like open source eventually made it into our everyday lives, TuringBots will, too.
How To Prepare
From our research of early adopters and suppliers of TuringBots, most assign a leader to kick off a pilot program to:
- Set boundaries on usage of TuringBots. Set clear rules to distinguish the use of TuringBots based on your different goals. Are you coaching and training neophyte developers? Onboarding developers for target projects and products? Or deciding if experienced developers are the only ones who can use TuringBots for production-ready development?
- Learn by experience and share best practices. Encourage sharing experiences in a repository as teams experience and experiment with TuringBots. Take note of how to write effective, reusable prompts that embed the best programming practices, postprocessing of large language model (LLM) outputs, security policies, or nonfunctional requirements.
- Review coding and architectural guidelines. Organizations should properly register and store all their code in a code repository. Better architectural documentation and coding artifacts will make it easier for TuringBots to learn your good practices. In turn, they will make good suggestions based on those good practices.
- Balance risk with use of TuringBots for production code. In the short term, your most experienced developers should have permission to leverage TuringBots to deliver higher-quality and safer code in production.
- Choose the right use cases. Assess your application and technology portfolio, evaluate the return on use cases, and compare risk. TuringBots are more mature in the custom coding world and less so in the programming packaged app space. “Plan,” “analyze,” “design,” and “dev-insights” TuringBots have a much smaller ecosystem than “coder,” “tester,” and “deliver” TuringBots, which means they are less mature.
- Spell out the risks of genAI TuringBots. If the TuringBot’s underlying technology is genAI, whether it’s a coder, tester, or deliver TuringBot, make sure your organization knows about the risks involved. Make sure your lawyers are checking how vendors are addressing IP protection, data privacy, security risks, copyright, and code attribution.
How We Can Help
The State Of TuringBots, 2023 and The Future Of TuringBots will give you a deeper picture of the TuringBots market now and in the next four to five years. Are you wondering what generative AI and TuringBots mean for your engineering or application development organization? As a technology executive, are you wondering what benefits and productivity gains to expect? If you are Forrester client, schedule a guidance session with me at email@example.com and I will help. If instead you want to share your experience with me on generative AI and TuringBots, email me at firstname.lastname@example.org.