News: 1713004212

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

75% of enterprise coders will use AI helpers by 2028. We didn't say productively

(2024/04/13)


Global tech research company Gartner estimates that by 2028, 75 percent of enterprise software engineers will use AI code assistants, up from less than 10 percent in early 2023.

As of the third quarter of 2023, 63 percent of organizations were piloting, deploying or had already deployed AI code assistants, said the survey of 598 software engineering leaders at large enterprises.

However, Philip Walsh, Gartner senior principal analyst, warns that there can be a mismatch between IT leadership's expectations and software teams' experience when it comes to productivity uptick.

[1]

He said vendors selling AI-assisted coding tools were claiming they could increase coder productivity by as much as 50 percent, while a third of CIOs (34 percent) and technology leaders thought AI code generation might be a "game changer" for their software development efforts.

[2]

[3]

"That's some really high expectations from the productivity gains from these AI code assistants," he said.

While the popularity of AI coding tools will undoubtedly increase, development teams might need to manage the expectations of their senior managers ... 'They're not going to hear it from the vendors'

But looking more closely at the vendors' claims reveals that the benefits of AI-powered coding tools can be confined to quite narrow tasks. For example, one study relied on an A-B style experiment where a team writing a web server in JavaScript was pitted against another that also employed AI coding tools. Writing a boilerplate for Python was another common comparison task, he said.

But these tasks might not be representative of the tool's abilities because of the abundance of training data online demonstrating how coders had already tackled the problems.

At the same time, coding itself did not make up the majority of effort in the whole software development lifecycle, Walsh said.

[4]

"There are a whole variety of tasks [involved in software development] from planning, design, research, and actual code generation and development … then a lot of testing and verification, and then deployment, configuration, and monitoring. Even if you're getting 50 percent faster task completion [on coding], that's only going to be 50 percent of 20 percent. So that means only 10 percent greater improvement in overall cycle time," he said.

[5]Forcing AI on developers is a bad idea that is going to happen

[6]Google's AI-fueled IDE Project IDX tries to show you how your app runs on Android, iOS

[7]IBM takes a crack at 'utility scale' quantum processing with Heron processor

[8]IBM says GenAI can convert that old COBOL code to Java for you

While the popularity of AI coding tools will undoubtedly increase, development teams might need to manage the expectations of their senior managers.

"They're not going to hear it from the vendors," Walsh said. "Hopefully, their developers and the engineering leaders are going to tell them and they're going to listen. What we do not recommend is some kind of top-down productivity mandate. That doesn't work."

He said development teams need the freedom to figure out the best use cases. They need to work in a learning culture of experimentation, with the freedom to fail to get the most from the new tools.

"CIOs need to create that culture and listen to their people, but also create that space for experimentation and failure," he said. ®

Get our [9]Tech Resources



[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/devops&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZhqsGPMR7sC2fuBLePFP8gAAAAg&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/devops&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZhqsGPMR7sC2fuBLePFP8gAAAAg&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/devops&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZhqsGPMR7sC2fuBLePFP8gAAAAg&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/devops&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZhqsGPMR7sC2fuBLePFP8gAAAAg&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[5] https://www.theregister.com/2024/02/12/opinion_column_on_forcing_ai_features_on_developers/

[6] https://www.theregister.com/2024/01/26/google_project_idx_cloud/

[7] https://www.theregister.com/2023/12/05/ibm_heron_quantum_processor/

[8] https://www.theregister.com/2023/08/22/ibm_says_genai_can_convert/

[9] https://whitepapers.theregister.com/



cantankerous swineherd

get your backdoors in the training data now, if you haven't already.

Let’s hope it improves

Ashto5

So far I have had to rewrite much of what was offered by these tools.

They will get better I am sure but honestly right now I am not worried.

Gartner estimates that by 2028, 75%

Mage

But if reality bites it could be 10%.

Unfortunately Senior Manglement listens to vendors and juniors, not experienced devs in their own company, because they are saying what Manglement wants to hear.

What's Gartner's track record on these predictions and are they using so-called "AI"?

mhoulden

I can imagine the remaining 25% will be trying to switch it off. It's bad enough trying to do something in Power Automate without Copilot popping up all over the place like a hyperactive Clippy.

Overhype

Filippo

Now Gartner is saying 'beware of high expectations'? Gartner?

First they came for the enterprise coders...

Will Godfrey

... but I was not one of those, so said nothing.

...

...

If only Dionysus were alive! Where would he eat?
-- Woody Allen