2023 was the year of generative AI, with applications like ChatGPT, Bard and others becoming so mainstream we almost forgot what it was like to live in a world without them. Yet despite its seemingly revolutionary capabilities, it's important to remember that Generative AI is an extension of “traditional AI”, which in itself is a step in the digital transformation revolution.
This means that in 2024, we’re likely to see businesses continue to seek ways to adopt generative AI as a way to enhance their operations. But this year, businesses will go beyond the hype. They will focus their resources on optimizing and adapting generative AI, and other AI technologies, attempting to turn them into a driving force for the business.
For data science practitioners, productization is key, just like any other AI or ML technology. Successful demos alone just won’t cut it, and they will need to take implementation efforts into consideration from the get-go, and not just as an afterthought. Considerations such as reducing and controlling potential risks, cost effectiveness, scalability, modularity and extensibility, and continuous operations must be part of any implementation.
What will be different in the way businesses approach generative AI this year? What are their expectations from this hyped technology? Will generative AI continue to be one of the hottest topics in 2024 as well? Here are my predictions for the upcoming year:
1. Beyond the Demo: From Prototyping to Generating Business Value
The excitement and attention surrounding generative AI is well-deserved, considering its potential and capabilities for both businesses and individuals. However, it's important to contextualize generative AI within the broader landscape of AI and ML technologies. Just like any other AI or ML technology, the real value of generative AI lies in its ability to create tangible business value.
Let’s look at an example of using generative AI to create a chatbot for a business. The aim should not be just generating conversations. Instead, these conversations should be tailored to fit the unique business context and target persona of your organization.
This requires addressing considerations like the company's brand, customer preferences, historical purchasing data, and potential recommendations for products or services. This contextual information, when combined with the capabilities of generative AI, creates a chatbot that is not only interactive but also highly relevant and meaningful for the business.
In 2024, the practitioners who will succeed are those that go beyond the demo and on to implementation and productization. This will require investing resources in the entire AI and ML lifecycle, including building the data pipeline, scaling, automation, integrations, addressing risk and data privacy, and more. By doing so, you can ensure quality and production-ready models.
2. Enhanced Focus on Accuracy
One of the current shortcomings of generative AI is accuracy issues, from hallucinations to suboptimal responses to lack of context and knowledge for certain verticals and topics. In 2024, I expect to see a growing focus on accuracy during AI model development and deployment. Businesses are recognizing the need to refine these technologies to ensure they deliver accurate, brand-consistent and high-quality results. While low accuracy is tolerated in pilot phases, it will not be overlooked in production environments.
Going back to our chatbot example, unlike informal ChatGPT usage, businesses cannot afford inaccurate, toxic, politically incorrect, or ungrounded answers. Various measures should be applied to guarantee the answers are exact and embody the brand's unique language, messaging, and target audience. A banking application will likely focus on strict answers, while a children’s playground app will focus on age appropriate content.
To meet these needs, practitioners will find themselves required to ensure that generative AI applications are accurate and reliable. Therefore, I predict this trend will drive significant changes in data pre-processing to guarantee data used by the application is accurate, not duplicated, and contains appropriate and risk-free data.
Models may need to be fine-tuned to follow the brand or application's unique attributes and voice. Extensive application testing measures and guard rails should be implemented. All aspects of the application, data, and model should be constantly monitored to verify accuracy and quality are met, and that there are no business risks, and costs don't jump beyond the allocated budget. This should all be done by following MLOps and CI/CD practices.
3. From All-Around Hype to Specific Use Case Realization
2023 was the year of generative AI hype. For those of you who are familiar with Gartner's Hype Cycle - a tool for measuring the maturity, adoption, and social impact of various technologies - generative AI likely reached the "Peak of Inflated Expectations."
I predict that in 2024, the narrative surrounding generative AI in the business world will go through a significant shift. Companies are beginning to recognize that this technology, despite its vast potential, is not a Swiss Army knife. Instead, it has specific strengths and limitations. This is mainly due to its technological complexities, high implementation costs and accuracy limitations.
After a year of enthusiasm (and perhaps overestimation) of generative AI's capabilities, businesses are now poised to gain a more realistic and practical understanding of where and how this technology can deliver the best ROI. This includes identifying the specific use cases where generative AI can provide the most value, thereby maximizing its benefits. For example, a customer-serving chatbot, applications that require text summarization, and more.
This realization will also likely lead to a broader enhancement of AI practices beyond just generative AI. Businesses will now explore additional AI technologies and methodologies that can complement or augment generative AI capabilities, creating a more comprehensive and effective AI strategy. This nuanced understanding of AI will encourage strategic integration of AI technologies where they can add the most value.
4. GenAI as a Productivity Force Multiplier (but not Necessarily the Main Course)
Generative AI is a powerful and innovative technology. However, it rarely operates effectively in isolation. Generative AI should be best understood not as a standalone solution but as a potent assistant and a valuable asset in the employee toolbox. Its true potential is realized when it is integrated into existing processes and workflows.
This integration can significantly enhance various types of work performed by the workforce, from code development to marketing strategies to the creation of presentations. For example:
- Development - Generative AI can automate repetitive tasks, suggest optimized code solutions, and assist in debugging and documentation.
- Marketing - GenAI can be used to generate content, analyze customer data for targeted campaigns, or predict market trends.
- Presentations and decks - Generative AI can create compelling, creative and sophisticated content and visuals tailored to specific audiences and based on vast amounts of data. This includes analyzing data and presenting it in infographics and charts, designing custom illustrations and building content around a cohesive narrative.
5. From Isolated Niche Tools to Integrated Solutions
The generative AI landscape is constantly evolving. Looking at a snapshot of the landscape three months ago reveals it is no longer up-to-date today, and today’s snapshot will be irrelevant three months from now. This is the sign of a vibrant ecosystem,
In 2024, I predict the industry will be moving away from these standalone applications towards integrated solutions, moving away from standalone applications. These will include a number of aspects:
- Out-of-the-box, vertical solutions - Solutions designed to be readily deployable or consumed as a SaaS offering, catering to specific industry needs without the requirement for extensive customization or development. This approach greatly simplifies the adoption of GenAI, allowing businesses to quickly implement and benefit from its capabilities.
- Embedding GenAI capabilities into existing software platforms - GenAI will become a standard feature, much like how advanced analytics are today incorporated into many business applications. For example, integrating GenAI functionalities straight into Salesforce or Office. Users of such platforms will benefit from GenAI's capabilities without needing to engage in separate development projects or integrate third-party tools.
This integrated approach has significant implications for the market dynamics. Large corporations, with their extensive resources and broad customer base, are likely to acquire or overshadow smaller entities especially if their products lack substantial technological differentiation. This potential for cannibalization by larger players suggests that merely developing a niche GenAI solution may not be a sustainable strategy in the long run.
After so many years in the AI and ML industry, it’s exciting to see companies attempting to turn AI technologies into the core of their business. AI has vast potential to help businesses find new revenue streams, cut operational costs, improve productivity, reduce friction and increase competitive differentiation.
By thinking about the ML process in advance: preparing, managing, and versioning data, reusing components, etc., businesses will be able to improve performance, reduce potential risk, save redundant engineering, and benefit from the full potential of AI.
On a personal note, I’d also like to share that my book “Implementing MLOps in the Enterprise: A Production-First Approach” co-authored with Noah Gift and published by O'Reilly is now available here.
Here’s to a successful 2024!