It’s been nearly a year since I first wrote about generative AI and it seems like not a day goes by where there isn’t a launch of a generative AI application. There’s a lot of excitement whilst people are exploring and developing strong use cases for generative AI. At the same time, there are concerns about the impact of generative AI on established disciplines like software engineering and music production.
Time will tell the true impact of generative AI, but it’s worth exploring ways in which it can benefit established disciplines. Let’s look at the impact of generative AI on three disciplines close to my heart: product management, software engineering and writing.
Generative AI can augment the role of the product manager, automating tasks that are valuable but time consuming: data analysis and prototyping.
Product Managers have access to a large pool of (customer and transaction) data. Analysing this data is often very manual and time consuming.
Product Managers can use generative AI to process and analyse customer data, generating valuable insights about customers and their needs. Tools like RapidMiner, BigQueryML and IBM Watson can help analyse and categorise customer data at speed.
Using cleaned up and analysed historical data, AI can predict user behaviour and market trends. Product Managers can use AI to simulate certain user behaviours and use data-driven insights to make product decisions. For example, you can use tools like Pecan and Microsoft Azure to analyse past customer buying behaviours to predict future buying behaviours.
In terms of impact on software engineering, the biggest promise of generative AI is that it will increase developer productivity. Like with many other disciplines, the expectation is that AI will help remove repetitive tasks and automate testing. However, a recent study by McKinsey highlights important limitations to AI (e.g. not knowing project or organisational context), and the need for human oversight when developing software.
Writing new code
End-to-end code generators like GitHub Copilot, Amazon CodeWhisperer, Tabnine and OpenAI Codex translate natural language into code. These tools can make it easier and quicker for software engineers to create new code, predicting and suggesting new code snippets.
Reducing the need for human intervention is the main benefit of automated testing. Testing tools that are augmented by AI automate common testing tasks. Think about tasks such as test case generation and test analysis where tools like Selenium, Katalon and Testim can speed up the process and reduce the need for human intervention.
Naturally, there are some important downsides to consider with respect to automated testing. Like with any software tool, the AI testing tools can fail to recognise an issue in the technology being a tested or incorrectly flag an issue. Or the AI is biased, which is a more generic AI issue. For example, where there’s bias inherent in the algorithm used to test the data it can render inaccurate information.
How much should authors worry about AI?! There already have been examples of deepfake books being sold on Amazon, and there’s real concern among authors about the impact of AI. There are, however, ways in which authors can use AI to their benefit:
Creating a narrative framework
AI tools like Subtxt, Jasper AI and Sudowrite help authors develop a structure narrative for their stories. For example, Subtxt’s “Narrative Agents” use a predictive narrative framework to develop a storyline. If authors get stuck on the outline of the story or struggle to develop an idea for a story, AI can suggest ideas for a structured story.
Authors can use AI to create and develop characters in a story. Through StoryFit, for example, authors can use AI to help authors figure out how many characters to include and analyse ways in which the characters can develop throughout the story.
Apart from helping with research, AI can be a useful aid for authors to help them improve the quality of their writing. Grammarly, ProWritingAid and Hemingway Editor analyse the quality of the written text, offering automated suggestions to improve grammar and style.
Main learning point: Whilst some of the fears and downsides about generative AI feel justified, it’s clear to see how established disciplines like Software Engineering can benefit from generative AI.
Related links for further learning: