5 Gen AI Developments That We Will See in Business in 2024
Expect more focus on ROI, productization planning, finops, code creation and customer support
By Premkumar Balasubramanian
New Year’s champagne bottles were still popping when generative AI made its first splash in 2023. Media far and wide reported that OpenAI’s ChatGPT reached 100 million monthly active users in January – just two months after its debut. Many of these stories also trumpeted UBS commentary, noting this made ChatGPT the fastest-growing consumer application in history.
Over the course of 2023, ChatGPT gained a reputation as a fun tool to play with. But ChatGPT and generative AI are so much more than that. Gen AI, which was a game changer in 2023 and a clear sign of the increasingly rapid pace of change, is revolutionizing the way that people work.
Gen AI will drive further change in business and how we work in 2024 and beyond.
Here’s how we expect things to play out.
- Businesses will work to get a handle on the ROI of gen AI
In 2024, look for more organizations to understand both the risks and what return on investment (ROI) they can expect from gen AI in terms of reducing the effort, time or number of people it takes to effectively handle particular tasks.
For example, a company may decide to feed AI with employee policy documents so it can more efficiently provide workers with details about benefits and company travel policies.
The company may do a proof of concept (POC), which could be completed in two or three weeks. But before it productionizes an application, it needs to make sure it has considered and taken steps to avoid hallucinations so that when an employee requests information about disability insurance, for example, gen AI will provide them with the right answer. Because if employees get the wrong company benefits details, they could get stuck with a decision that they wouldn’t have made if they had a concrete benefits policy document in front of them. As generative AI is a subset of AI that focuses on generating new content, it is really imperative that companies focus on understanding the hallucination risks in use cases like the above.
The work of productionizing POCs will involve a fair amount of effort and time. A lot of that work will take place in 2024, and that is when the money will start to flow.
- Companies will seek help to determine the best AI models for them
In the year ahead, there will also be consulting services designed to help organizations understand what form of AI is the right AI for their particular needs and specific applications.
Not everybody needs the top-of-the-pack GPT-5, if it comes out next year. To determine which AI model is the best match for them, businesses will work with AI experts that provide advice based on their specific use cases to help them keep costs low in the long run.
Addressing costs upfront is important because, for POCs, the cost differentials between various AI models are minimal. But as you scale the model to thousands or millions of calls across users, you will pay a hefty price because these models run on costly compute and high-end storage.
- More organizations will embrace finops for gen AI and AI observability
FinOps for gen AI will help organizations address cost management once they roll out gen AI because the costs stem from more than the API that you call. Businesses that use gen AI also need to consider the costs of the ancillary services they will be using and need to pay for.
If a company feeds an AI model with all of it policy documents, as discussed earlier, and it wants to have a local report, then it will need the storage and compute for the local report. The company may also need a vectorized database. The database could be local or it could be in the cloud, but wherever that database may live, the company will still need to pay for the software.
AI observability products are also poised to see greater adoption in 2024 to ensure explainable AI, keep toxicity low and help companies understand AI bias. More businesses are also likely to adopt hallucination-free products. Once use cases go into production, running an ecosystem efficiently – from implementation to production to evolving as new services emerge – is critical.
Responsible AI (RAI), which is the process of adhering to ethical and social practices while developing and deploying AI systems, is another important aspect of AI that companies will focus on more going forward. Look for more startups to emerge in the RAI space.
- Software engineering teams will use gen AI far more in 2024
Using AI to generate code will become more prevalent next year.
Of course, everybody already knows that gen AI can generate code. The challenge is: If you use gen AI to generate code, how can you ensure there are no copyright infringement issues you will run into when you want to productionize the product? If you use AI-generated code, you want to avoid setting yourself up for someone to come after you for copyright infringement. But as organizations resolve these technicalities and legalities around usage, more software engineering teams will adopt co-pilots to support development and enhance productivity.
Goodwin Law partner Robert D. Carroll writes that businesses can mitigate the risk of copyright infringement in such situations by getting a license or a representation and warranty from the provider of the gen AI tool they are using to ensure that the source works on which the tool is trained are licensed and that the license extends to the user; running a source code audit program to assess whether gen AI-created code is similar to other existing code and then taking steps to comply with the relevant open source license or excising the similar code; and conducting due diligence on the gen AI tool provider to understand what source materials it uses.
- Gen AI will play a growing role in customer support and success
Many organizations, including Hitachi Digital Services, are also experimenting with gen AI in customer support and success. Capturing and cataloging support interactions can enable AI to understand what issues are most commonly reported and how one solution can be applied to another.
That’s very different from the common approach today, in which customer service agents have to type everything into their systems, there is some text correlation, but if the keywords don’t match, the issue will be perceived as a different problem because there’s no contextual understanding of that data. However, if a system can capture and analyze entire conversations and the exact words spoken, AI can generate lists of common issues, common solutions and common takeaways that can be used to enhance support and improve customer experience.
AI and neural networks have been around a long time. (I did a college project on these topics 30 years ago.) But in the past year, we’ve all experienced the ease of access that gen AI brings. Now, with gen AI, everybody can talk AI – people don’t need to be AI experts to use AI.
In the year ahead, gen AI is going to pop up even more widely and in more use cases.
And your business is going to need to be ready for these changes to remain competitive.
Premkumar “Prem” Balasubramanian is SVP and CTO, Digital Solutions at Hitachi Digital Services, where he is responsible for strategizing and supporting all go-to-market pursuits, offering shaping, architecting repeatable solutions and providing technology and thought leadership in the areas of cloud, data, IoT and gen AI. Prem joined Hitachi from Cognizant, where he spent over 18 years as a technology and practice leader, focused on helping customers with their transformation and modernization journey. During his tenure, Prem assumed a variety of leadership roles with the most recent being the head of Cognizant’s Digital Architecture and Resilience and Reliability Engineering Practices. Prior to Cognizant, he worked in technology roles at RippeTech LLC and California Software Labs in a variety of areas including device driver development, network sniffers and networking protocols. Prem is passionate about building always-on IT systems and holds a patent on “systems and methods for supporting resilience in IT environments.” Prem earned a bachelor’s degree in computer science and engineering and a post-graduate diploma in business administration.