ChatGPT's AI revolution has enterprises rethinking their IT automation strategy
The sudden breakout of ChatGPT since its launch in late November has marked the dawn of a new AI era. With Microsoft and Google officially in an AI arms race, many enterprise IT leaders are being forced to rethink their approaches to AI and automation. With many potential use cases across their corporations and within their IT departments, they have to carefully consider where best to implement this technology, along with all the pros and cons.
Generative AI is not limited to predicting words or powering Bing's chatbot; it also shows promise for automating many engineering and development tasks, which provide major productivity benefits. However, if AI is used for code generation, there is no telling where it could borrow prewritten lines of code from, which quickly raises intellectual property red flags. Worse yet, the more days we spend using these new AI tools, the more we find they offer easy answers, which are often not the right ones.
Over the last few months, I've talked with many enterprise leaders across Fortune 500 organisations about the advancing AI wave and its impact on automation – here's my current advice to anyone rethinking their strategy, either within their engineering teams or beyond.
Humans, not AI, should be doing the intelligent work
While AI advancement is moving faster than ever, AI isn't that intelligent yet. In tests we've done internally at Adaptavist, using ChatGPT to develop code, it produced something machine-readable, but that failed to function as software. So it's clear we still need the human factor to continue producing intellectually-intensive, critical work, such as writing code, within most IT departments today.
But that isn't the only reason we advise corporations to focus AI and broader automation efforts on tedious, repetitive tasks and leave the thinking to their employees. Another big factor is employee engagement. If enterprises want to keep workers engaged and encourage retention, they need to provide them with challenging, meaningful work. As my organisation's 2022 Digital Etiquette: Reinventing Work Report found, of the one-third of UK workers actively searching for new job opportunities last year, 26% noted that they were seeking a new job with more purpose.
And whilst over 20,000 technology workers have been laid off from UK jobs in 2023, the competition for talent in the industry remains. For instance, staffing provider Manpower's recent survey of employers highlighted that in the IT sector, 76% of hirers said it was still challenging to find people with the necessary skills to fill open positions. Therefore, some strategic enterprise IT leaders are using AI and automation not to replace workers, as many have feared and Mckinsey famously predicted several years ago, but as a tool to eliminate repetitive tasks and give employees more time to focus on meaningful work and creative, higher-value activities. The initial results are higher productivity and happier employees.
Using tech tools to get human colleagues working better together
Implementing enterprise automation impacts people, processes, and tools, but most enterprises fixate on the tools and processes while overlooking their people. Enterprises that can illustrate internally that customer value creation from automation will come from augmenting employees' abilities and streamlining human collaboration will create a competitive advantage. That's why, in addition to giving employees their time back, we're also assisting IT leaders interested in using AI and automation to help their employees work better together.
For example, while Microsoft's use of ChatGPT within its Bing search engine has captured most of the attention, it may actually be the ways it has embedded AI into Microsoft Teams that hold better foresight into where AI can be the most beneficial in this capacity. In early February, Microsoft announced that the AI from ChatGPT was being used to execute chores such as taking notes and bulleting takeaways from the Microsoft Teams video conferencing tool. In the modern distributed workplace, this process automation can effectively get human colleagues to work better together — using technology to fill gaps in communication.
We've seen the same thing with our Fortune 500 customers and how they use ScriptRunner within development teams reliant on Atlassian's JIRA. As one of the most popular work management tools for software teams, JIRA is used by IT teams to track, organise and prioritise engineering issues. One major financial organisation we work with has been using the automation benefits of the ScriptRunner plugin to ensure their IT support team is equally sharing the workload to speed up the issue triage process. They now automate the flows for new IT support tickets by quickly writing a script. Suppose one person on the support team is handling 15 tickets, and another is working through 5 tickets. In that case, ScriptRunner will automatically bring a new ticket to the support staff with less of a current workload.
While this may sound like a basic application, simple automation can create more value than complex automation, given that it has more reusability within the entire enterprise. You could imagine the same use case above for customer support and sales teams. IT should lead this standardisation of automation patterns and tools that can be implemented across the IT department and the entire organisation. And process automation use cases like these that augment human tasks and make human colleagues work better together can be replicated across enterprises.
Automating code review and testing
While it's currently too risky to rely on AI to write new code, AI does show significant promise in analysing bugs and errors in lines of code at scale. Bugs and errors remain a major issue that needs to be addressed, as some reports indicate that up to 50% of IT budgets are spent on debugging. Intel is one company that has looked to address the challenge with its solution ControlFlag. The solution utilises machine learning to detect problems in computer code — and can greatly reduce the time required by IT professionals to debug software.
Some teams might need more time to be ready for AI-powered code review. However, that doesn't mean that automation can't still play a role in streamlining the process. Internally, our team has relied on automation to simplify a peer code review system we have in place. At the beginning of each engineering sprint, every developer is automatically assigned a human review partner. That means twice a month, workers are given a new colleague to review their code. Before adopting our rotating peer reviewer system, it could be challenging to manage who should check each pull request. To remove the manual process, we set up some simple automation that matches code reviewers twice a month and even sends a Slack message when it has completed the action.
The testing process also shows much promise for AI and automation. A global 2022 survey of 2,600 DevOps practitioners by Tricentis found nearly two-thirds of respondents cited testing as the area within DevOps where AI will have the most impact. This type of automation can fuel continuous deployment, where testing is done rapidly, and software can be moved to production with limited delay.
Of course, as automation moves further into code development and deployment, we'll see automation in the form of low and no-code solutions. As these tools improve, we could see a marked shift in team composition, with non-developers playing a more prominent role than ever before in historical IT tasks. Automation here will enable enterprises to expand their technical capabilities without hiring more developers. It will be another new way to shift functions historically reserved for IT across departments, freeing developers' time for more strategic work.
One thing that can't be argued is that AI and automation will drive continuous change over the next decade within enterprise IT departments. IT leaders who master transformation management of their people, processes, and tools will be the ones that not only survive, but thrive long-term.