What kind of future will AI bring enterprise IT?

If you’re a business looking for ways to ride out a recession harder while beating your competitors in the process, open source isn’t the answer. Neither does the cloud. It is true that both can be useful. Both are ingredients of how companies need to rethink their traditional IT approaches. But neither will do much to set you apart.

Because? Because everyone else is already using open source and the cloud too. There was a time when being an early adopter of the economics of open source projects like Linux or MySQL could set a business apart, but not anymore. Enterprise cloud adoption is still nascent (approximately 10% of all IT spending by 2022, according to Gartner estimates), but adoption is advancing at such a rate that it probably won’t distinguish your customer experience alone through the cloud. What will set you apart?

Machine learning (ML) and artificial intelligence (AI). But maybe not as you think.

Think progressively about AI

This is not one of those articles that touts AI/ML as an ill-defined panacea. Yes, AI and ML have been instrumental in developing powerful drugs to fight COVID-19, and could even one day help find a cure for cancer. But there’s no magic AI/ML fertilizer that you pour into dying IT projects and they magically flourish. Companies like Google or Uber have been at the forefront of AI/ML, but let’s face it: you don’t have their engineering talent.

Even these companies are using the recession to spend less time on trips to the moon and more time on incremental advances, as a recent article in The Wall Street Journal (“Big Tech Stop Doing Stupid Things”) Calls: The tech industry “that has worked long to disrupt is now focusing on improving what’s already there.” Instead of reinventing the wheels, the article notes: “The best technology investments of 2023 could be companies content to spend their money on greasing [the wheel].”

A big way companies are doing this is with AI/ML, but not gee-whiz flying cars. AI/ML is being used in much more pedestrian (and useful) ways.

Zillow spent years trying to use AI/ML models to do big house flips. However, at the end of 2021, the company exited that business, citing an inability to forecast prices despite sophisticated models. Instead, Zillow has gotten pragmatic and is using AI/ML to help prospective renters view listings as they walk through town and allows owners to build floor plans from photos of those apartments. Much less sexy than a billion dollar home remodeling business, and much more helpful to clients.

Google, for its part, has begun retailers that offer the ability to track store inventory by analyzing video data. Google trained its models on a dataset of over a billion product images. It can recognize the image data whether it comes from a mobile phone or a store camera. If it works as advertised, it would be a boon to retailers who have traditionally struggled to control inventory. Not a sexy use of AI/ML, but useful for retail clients.

Microsoft, the leader in AI/ML, has just made a big investment in OpenAI, with the reported intention of bringing GPT-esque functionality to its productivity applications, such as Word or Outlook. Microsoft has the resources to bet big on an Office moon makeover, perhaps making it entirely voice-powered. Instead, it’s likely to give Office a major Clippy update with a GitHub Copilot kind of approach. That is, GPT could take over some of the undifferentiated drudgery of writing documents or creating spreadsheets. Less sexy, more useful.

Choosing not to miss with AI

The incremental approach turns out to be the smartest way to build with AI/ML. As AWS serverless hero Ben Kehoe argues, “When people imagine integrating AI…into software development (or any other process), they tend to be overly optimistic.” A key flaw, he stresses, is believing in AI/ML’s potential to think without commensurate ability to fully trust its results: “A lot of the AI ​​shots I see claim that AI will be able to take on all the responsibility for a given task for a person, and implicitly assume that the person responsibility because homework will just… evaporate?

In the real world, developers (or others) have to take responsibility for the results. If you use GitHub Copilot, for example, you are still responsible for the code, no matter how it was written. If the code ends up with errors, it won’t work to blame the AI. The person with the pay stub will be at fault, and if he can’t verify how he arrived at a result, he’s likely to scrap the AI ​​model before he quits his job.

This is not to say that AI and ML don’t have a place in software development or other areas of the business. Just look at the examples from Zillow, Google, and Microsoft. The trick is to use AI/ML to complement human intelligence and allow that same human intelligence to verify the results. As Kehoe suggests, “When looking at claims AI is going to automate some process, look at what the really difficult inherent complexity of that process is, and whether the process would be successful if a high degree of (new) uncertainty [through black-box AI] he was injected into that complexity.”

Adding uncertainty and making accountability more difficult is not a start. Instead, companies will look for areas that allow machines to take more responsibility and the people involved to be accountable for the results. This will be the next big thing in enterprise IT, precisely because it will be so many small, incremental things.