Home The Positive and Negative Impact of AI & Automation on Software Developers

The Positive and Negative Impact of AI & Automation on Software Developers

by Ant Sh
659 views
The Positive and Negative Impact of AI & Automation on Software Developers

As more companies seek to adopt artificial intelligence foundation models, such as OpenAI LP’s ChatGPT, it’s becoming increasingly clear that automation platforms present a potential problem for software developers.

On the one hand, these platforms can potentially revolutionize how developers work. On the other, they also threaten job security for those in the industry, presenting a stark contrast to established processes.

That said, there’s no question platforms such as ChatGPT are a disruptive force and have been enthusiastically embraced by venture capitalists. Such firms have poured more than $1.7 billion into generative AI solutions over the past three years, with a significant portion going toward AI software coding, according to research firm Gartner Inc.

The impact of automation on software development has other potential implications, especially regarding the quality and diversity of code produced.

“Large language models can reduce complexity and accelerate the adoption of enterprise automation platforms,” theCUBE industry analyst Dave Vellante wrote in a recent edition of his Breaking Analysis series. The flip side is that software robots are designed to improve human productivity through intelligent automation, and GPT models could cannibalize some, if not many, use cases initially targeted by RPA vendors.

The new reality is causing customers to rethink their automation strategies and is pushing vendors to evolve their messaging rapidly to position foundation models as an accelerant to their platforms.

How might this new technology change open-source developer communities? How might emerging automation platforms represent a double-edged sword? And, in the end, how might this change the industry forever?

The real-world applications of AI

Before the explosion of AI, automation was already moving at a breakneck pace. While automation and technological advances are predicted to displace 85 million jobs by 2025, 97 million new roles will be created “as humans, machines, and algorithms increasingly work together,” the World Economic Forum recently predicted.

Open-source generative AI, such as GPT-3, a natural language processing model developed by OpenAI, will likely disrupt most technology fields, and software developers are not immune.

It could, for example, be used to automate code generation — something that some startups, such as Hugging Face Inc., have already started preparing for.

“Hugging Face is a resource for startups and other businesses around the world. Our transformers can help them build virtually any natural language processing application at a fraction of the time, cost and complexity they could achieve on their own, helping organizations take their solutions to market quickly.”

Clement Delangue, Chief Executive, Hugging Face

Others, such as the open-source repository GitHub, view AI-powered auto-completion as “just the starting point.”

“We’re testing new capabilities internally where GitHub Copilot will automatically suggest sentences and paragraphs as developers create pull requests by dynamically pulling in information about code changes,” said Thomas Dohmke, CEO of GitHub, in a recent blog post.

AI technology has already been utilized for code optimization, suggesting improvements and reducing the likelihood of errors. It’s also being used to automate testing. Still, even though ChatGPT has created a “Netscape-like” moment in the technology, as Vellante puts it, it’s not yet clear how IT decision-makers will directly implement it in their organizations.

As part of a recent Breaking Analysis, Vellante included direct quotes from a recent roundtable of chief information officers. When it comes to robotic process automation, machine learning and AI, one CIO said all three were essentially trying to solve the same business use case — removing excess resources, human or otherwise.

“What OpenAI has shown with ChatGPT is that you can get rid of a lot of ‘overhead,’ complicated artifact-building around typical RPA,” the CIO said. “I see that as a very interesting value proposition — to be able to supplant some of these workbenches in ‘classical’ RPA that take quite a while to master and quite a while to get any value past the regular use cases.”

Others have sought to make such technologies more accessible. Red Hat Inc. has sought to create community-driven AI as part of its Project Wisdom initiative, with an eye toward simplifying automation in infrastructure.

“This is the beginning of the community journey now. We’re going to start working together through channels like Discord and what not to be able to exchange information and bring people in.”

Tom Anderson, VP and GM for the Ansible Business Unit at Red Hat

The concerns to human developers

The flip side of the issue involves how this emerging technology could potentially replace human developers entirely. Indeed, companies such as Goldman Sachs Group Inc. have already started experimenting with generative AI internally to assist with code writing.

There are other concerns should organizations rely too heavily on code generated by AI, such as fragmentation if different foundational models emerge for various tasks. That could make it difficult for developers to collaborate effectively. Then, there could be issues if checks and balances aren’t in place for code generated by AI, especially given that ChatGPT today still has the potential for what has been called “hallucinations.”

“It will tell you very convincingly what it ‘thinks’ is right, no matter how wrong it is,” said Henrik Roth, co-founder and chief marketing officer of neuroflash GmbH. ChatGPT is a great tool for creative writing and advertising, Roth said, but “if facts matter more — such as in journalism and science — one should fact-check every claim.”

How companies see the road ahead

Automation platforms will inevitably change the game for software developers. The big players, especially those at the intersection of machine learning, AI and automation, are using RPA as a proxy, including Microsoft Corp., Amazon Web Services Inc. and Google LLC.

“Microsoft is well-positioned to capture wallet share, and observers can expect its relationship with OpenAI will be a linchpin of its AI strategy. Microsoft was arguably behind AWS and Google in AI from a technology standpoint but appears to have jumped to the lead from a business-model perspective,” Vellante said.

As part of his Breaking Analysis, Vellante also included results from Enterprise Technology Research, who asked customers if they were evaluating GPT models and for what use cases. Aside from a surprising 56% of customers who said they were not evaluating, the second-highest number were those who were evaluating it for customer chat.

“At the surface, one would conclude that RPA and automation platforms could benefit from GPT models and that these use cases are largely complementary. For example, a foundation model could write or accelerate the development of automation code that could direct software bots,” Vellante wrote. “But at the same time, there’s a looming overlap between the capabilities of large language models and some of the early RPA use cases; and this overlap will most likely grow over time.”

What comes next?

Both opportunities and challenges loom for developers, as AI foundation models have the potential to revolutionize how work is done, even if a threat hangs over job security and established processes.

The market remains a “tale of two cities”: Tough times mean companies look for ways to cut costs and where automation can come into play, in Vellante’s view. At the same time, he said that companies must spend money to make money.

“GPT models are catalyzing new thought, and both buyers and sellers are pivoting to turn foundation models into opportunities,” he said. “Initial use cases for GPT models are interesting, but not a direct replacement for enterprise automation platforms. However, low-end automations are at risk, and there’s no question there is a Venn diagram here intersecting foundation models and automation platforms.”

Indeed, generative AI could cannibalize some RPA use cases. On the other hand, the two technologies could work together to automate a wider variety of tasks.

“Nonetheless, all vendors in our view must leverage GPT models to simplify and accelerate adoption; and buyers have to step back and do some experimentation to see how they can deploy these new innovations to add value to their business.”

Dave Vellante, theCUBE

Similar to most technologies being upended by advancements in AI, automation platforms are changing forever. Only time will tell what they’ll look like on the other side.

The Source