Game changer:

The startling power generative AI is

bringing to software development
A newly emerging type of artificial intelligence promises
to rewrite the way software is built and maintained
Software development is poised to take a dramatic leap—with big implications for how developers work and
how corporations build and maintain the digital code underlying their operations.

The catalyst for this dramatic transformation is generative artificial Exhibit 1: Racing to one million ChatGPT hit one million users in
intelligence (AI), an emerging form of AI that can create original articles, record time
essays, images, music, and yes, code, by building on patterns it finds in
existing text, audio files, images, and software. Netflix 3.5 years
Generative AI has already gained notoriety for its ability to produce images
and text. Late last summer, it generated headlines when an AI-generated Facebook 10 months
image won an art contest. In November 2022, it captured the world’s
attention again with the debut of ChatGPT, a wickedly sophisticated, next-
Spotify 5 months
generation chatbot created by OpenAI, a research lab in San Francisco.
Users order up new content by giving these tools simple prompts like “how
did the Space Race change America?” (text) or “a purple rabbit hopping Instagram 2.5 months
on orange clouds” (image). ChatGPT attracted more than a million users in
just five days, making it one of the most rapidly adopted new technology
ChatGPT 5 days
platforms in history. (Exhibit 1)

Source: DiploFoundation, Generative AI models - a fun game that can easily get out of hand? December 22,
2022. https://www.diplomacy.edu/blog/generative-AI-models-a-fun-game-that-can-easily-get-out-of-hand/

© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent Game changer: The startling power generative AI is bringing to software development 1
member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
A paradigm shift as significant as the cloud or DevOps
While the public plays with this new toy—and college professors try to Skeptics are plentiful, to be sure. Some users question generative AI’s
figure out how to differentiate ChatGPT essays from student work—similar capabilities. It is a nascent technology whose output can sometimes
platforms are being used to generate software, including GitHub Copilot, miss the mark—or simply be incorrect. But in our view, much of today’s
also based on OpenAI models, and Tabnine, from the Canadian company skepticism is misplaced. The biggest challenge to collaboration between
of the same name. These systems have kinks and won’t put developers developers and generative AI isn’t the technology’s capabilities but its
out of business anytime soon. But they are making cost for organizations that want to work not just with
it possible for developers willing to work with their open-source code but also with the source code and
limitations to hand off the repetitive parts of their technical documentation the company uses to run its
jobs, supercharging their productivity and speeding For many own operations. Training a generative AI model this way,
the time to deployment for new software.
To reiterate, this isn’t a few years away. As we
developers, either on their own or with the help of a consultant,
could prove an expensive undertaking.
write, generative AI models trained against the
vast expanse of open-source code available on the
generative AI will Still, what generative AI is already doing is so
compelling that it’s hard to believe companies won’t
internet are already explaining poorly documented
code, generating documentation for code, and even
become the most keep pushing forward. Take OpenAI’s DALL-E, one of
several generative AI models. It can produce high-quality
writing functions or relatively targeted pieces of valuable coding images—art—based on natural language descriptions
code, all with minimal direction from humans. For
many developers, generative AI will become the partner they will (prompts) entered by users. Like ChatGPT, it’s based on
a third-generation “Generative Pre-trained Transformer,”
most valuable coding partner they will ever know.
If it lives up to its promise, generative AI will herald
ever know. GPT-3, from OpenAI. GPT-3 features a deep learning
neural network with 175 billion machine learning
parameters. OpenAI allows limited tinkering with this
a paradigm shift every bit as significant as the cloud
model for free, but charges for more extensive work.
or DevOps, the development process that enables
faster delivery of more reliable software products and services. In terms More directly for this discussion, look what’s happening with GitHub
of how corporations develop and maintain software, it will prompt changes Copilot, which uses the OpenAI Codex artificial intelligence model to
as big as, and likely even more impactful than, those created by Agile suggest software code and entire functions based on natural language
development methods, which enable rapid responses to changing software prompts. It’s already affordable in many applications. The cost for individual
requirements and customer feedback. programmers to use it is $10 a month. For businesses the cost is $19 per
month per user.1

1
Pricing for GitHub Copilot for Business, https://docs.github.com/en/enterprise-cloud@latest/billing/managing-billing-for-github-copilot/about-billing-for-github-copilot

© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent Game changer: The startling power generative AI is bringing to software development 2
member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
Still, for chief information officers and their employers, the most valuable This would be especially compelling in weak economic environments when
near-term use case for generative AI is likely to be the previously mentioned companies are faced with added cost pressures, as it would allow them to
training of it on the proprietary source code and technical documentation get more productivity out of their existing software development team. It
being used by their own organizations. “Fine-tuning” generative AI models would be particularly useful to those with the most intransigent technology
this way could lead to huge productivity gains and change forever the stacks.
economics of creating and maintaining software in a corporate environment.

Among other things, we see developers at forward-thinking organizations

turning to generative AI models to:
Generate unit tests that can help reduce human error in
1 Explain the functionality of code they did not write but are
now tasked with maintaining or upgrading. 6 coding and resolve corner-case errors that occur outside normal
operating parameters.

2 Provide guidance on which classes or libraries of code have

the necessary functionality to implement specific requirements. 7 Translate code from one language to another, helping with
legacy migration and development of use cases.

Improve the consistency of code through automated

3 refactoring, style changes, and automatically generating
explanatory comments.
8 Write documentation such as user stories and success
criteria.

4 Simplify code by suggesting less complex alternatives that can

ease future maintenance. 9 Accelerate debugging with generative AI’s ability to quickly
identify coding mistakes.

5 Explore alternative solutions to coding challenges, widening

the developer’s perspective on possible options. 10 Write code that goes beyond templates or stub code to
jump-start new tasks.

© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent Game changer: The startling power generative AI is bringing to software development 3
member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
Do things that were previously impossible

By lowering barriers to entry for new developers on complex codebases, generative AI also will allow
companies to do things that were previously impossible, such as:

• Rapidly onboard large groups of developers to accelerate new • Use offshore resources more effectively. A contextually informed
features or major changes to software. These developers would be generative AI assistant could provide guidance to an offshore coding
productive quickly and require less guidance from existing developers. team while senior onshore engineers and product owners get some
sleep. This also may enable changes to the resource mix to take greater
• Make developers more fungible across frameworks, platforms,
advantage of offshore resources.
products, and systems of record. Generative AI will provide the
scaffolding and guidance they need to work on a wider range of projects The level of effort required to fine-tune a commercial generative AI model
than they would normally be able to handle. on a corporation’s proprietary codebase, and the efficacy of that fine-tuning
on specific commercially available generative models, remains unknown at
• Make corporations less beholden to developers of their legacy
the moment. It is possible that the effort, and associated cost, will prove
codebase as more of that legacy knowledge is built into a fine-tuned
beyond the means of some organizations, at least in the near future. But
generative model.
the potential value is so great that we believe any large organization with
• Make the purchase or sale of software-related intellectual property a lot of developers that isn’t urgently exploring the opportunity is doing its
less dependent on the retention of developers. Generative AI will be stakeholders a disservice. A proof of concept at any large corporation with
able to codify much of their legacy knowledge. An existing fine-tuned a significant legacy codebase would go a long way toward demonstrating
generative AI model also would improve the quality of technical due the technology’s potential—and likely give that organization a leg up on its
diligence during the acquisition. competitors.

© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent Game changer: The startling power generative AI is bringing to software development 4
member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
Remaining challenges and risks
As with the adoption of any new technology, a broad move to use In the meantime, intellectual property issues around generative AI remain
generative AI in the corporate environment will surely hit some bumps unresolved, too. These models are trained on open-source code, with many
along the way. For all of their sophistication, current iterations of generative different types of licenses, and it remains to be seen what will happen if
AI can still generate flawed results. Even where results are rock solid, software they generate is deemed too similar to open-source code.
building in traceability and transparency may be important because many
Residence considerations for intellectual property are a potential sticking
who come to rely on the technology will not fully understand how its
point, too. A large bank concerned about trade secrets, for example, may
outputs are achieved—and, as such, not trust it. It will be important for
be reluctant to share its data with a start-up generative IT vendor operating
organizations to identify risks early and understand how those risks may
in a multitenant cloud environment. This could create a market for a larger
impact building trust in the use of generative AI.
vendor—perhaps Microsoft, Oracle, or IBM—to offer a single-tenant product
Overcoming management skepticism could be a particular obstacle. It more sensitive to those concerns. Microsoft, which invested in OpenAI
may be hard to get senior executives to believe developers will not blindly in 2019 and again in 2021, has since announced an additional multiyear,
follow their AI models and leave their organizations open to potential multibillion-dollar investment into the organization.
missteps triggered by what the executives may see as black-box technology
As noted, costs will be a stumbling block for some organizations over the
platforms. Developers and their bosses will need to sell senior leadership
near term, especially those seeking access to state-of-the-art generative IT
not only on the technology but also on the safeguards to be built around it.
models like OpenAI’s GPT-3.
All that said, generative AI will inevitably become more accurate and reliable
Ultimately, we see generative AI becoming an invaluable addition to the
over time. What’s more, the entire software development lifecycle process
advanced analytics and AI toolbox. However, organizations will need to
in a modern enterprise is already designed to find and fix bugs caused by
develop governance protocols around the responsible use of AI. The aim
mistakes developers have made. Errors attributable to generative AI can be
will be to ensure the technology is used in line with their standards, with
mitigated by the same processes. With additional training of the technology,
consideration of the impact on customers and society, and to build trust in
along with monitoring to ensure that outcomes align with expectations, we
AI-generated outcomes.
believe businesses will build trust in its capabilities.

© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent Game changer: The startling power generative AI is bringing to software development 5
member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
 The importance of getting started now
Although already impressive, generative AI technology is advancing rapidly.
Another large leap is capabilities is expected within this year. (Exhibit 2)
Even if organizations deem the technology insufficiently mature in its
current state, doing proofs of concept now to become familiar with the
technology is prudent and will help position them to respond quickly when

How KPMG can help

more advanced models are released. With all of the challenges and risks
posed by generative AI, the long term upside still remains extremely
positive.
An early and enthusiastic advocate for the power
Exhibit 2: A fast ascent The power of generative AI models is of artificial intelligence, KPMG is positioned to help
increasing at a rapid rate. Graphic shows model size in billions of your organization leverage generative AI to transform
parameters for key generative AI models introduced since 2018. software development and maintenance. Drawing on
our deep expertise in machine learning and natural
1000 language processing, we can guide your organization
through strategy development, platform selection, and
GPT-3
Model size (in billions of parameters)

Megatron-Turing implementation—and then provide ongoing support to

100 (175B) NLG help you optimize your investment in this transformative
(530B)
technology. We understand both the promise of generative
Megatron-LM Turing-NLG
10 AI and the process and cultural changes—including the
(8.3B) (17.2B)
T5 embrace of responsible AI practices—that will be required
(11B) to realize its full potential.
1 GPT-2
(1.5B)
BERT-Large
0.1 (340M)
ELMo
(94M)
0.01
2018 2019 2020 2021 2022

Chart courtesy of Nvidia. Source: Nvidia Developer Technical Blog, Using DeepSpeed and Megatron to Train
Megatron-Turing NLG 530B, the World’s Largest and Most Powerful Generative Language Model, Paresh
Kharya and Ali Alvi, October 11, 2022, https://developer.nvidia.com/blog/using-deepspeed-and-megatron-to-
train-megatron-turing-nlg-530b-the-worlds-largest-and-most-powerful-generative-language-model/

© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent Game changer: The startling power generative AI is bringing to software development 6
member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
Contact us
Mark Shank Kelly Combs
Principal, Advisory Director, Leader of Responsible AI
Lighthouse Lighthouse
KPMG in the U.S. KPMG in the U.S.
mshank@kpmg.com kcombs@kpmg.com

Some or all of the services described herein may not be permissible for KPMG audit
clients and their affiliates or related entities.
The information contained herein is of a general nature and is not intended to address the circumstances of any particular individual or entity. kpmg.com/socialmedia
Although we endeavor to provide accurate and timely information, there can be no guarantee that such information is accurate as of the date it is
received or that it will continue to be accurate in the future. No one should act upon such information without appropriate professional advice after
a thorough examination of the particular situation.
© 2023 KPMG LLP, a Delaware limited liability partnership and a member firm of the KPMG global organization of independent member firms
affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved.
DASD-2023-11580