As we launch into 2023, the world is in the middle of three very large transitions. First is the transition to a new geo-political, social-economic work order. Second is the transition to a low-carbon economy. And third is the first truly scaled deployment of AI in the enterprise. And across these opportunities and challenges, a few key technology trends are clearly taking shape:
1. Enterprises are moving from data-poor to data-rich
2. Transformation efforts will shift from big data to big AI
3. Digital transformation becomes 80% people, 20% tech
1. Moving from data-poor to data-rich enterprises
Enterprises are shedding their legacy of being data-poor and becoming data-rich companies using data to fundamentally drive their businesses. Two reasons have primarily driven this – data as a new asset class and new AI capabilities that make data more usable.
Data has emerged as the top driver of transformational value. As a result, we increasingly view data as its own asset class, as opposed to a byproduct and part of automation. It’s no surprise why. For example, in travel, airlines that provide a single application across schedule checking, travel booking, seat selection, flight check-in, baggage tracking and loyalty program management are delighting customers and differentiating their services. This drives an urgency to connect entire customer value chains across the data that sits in silos and predict and recommend next best actions that drive business results. As a result, data leaders now have a voice and a clear seat at the table, and boards and CEOs are driving a new data-driven business culture in their enterprises.
In parallel, we have new advanced toolkits lighting up massive amounts of dark data, making it productive and usable. Companies have always had information - but data comes in various forms, including structured data, such as transactional data, and unstructured data, such as social media posts, making it difficult to integrate and analyze – fundamentally data has been “dark”. AI is now allowing us to light up this data by extracting it from unstructured files like pdf documents, normalizing and cleaning it, and making it easy to access and use. For example, AI systems now routinely ingest hundreds of thousands of financial filing documents and balance sheets and automatically extract key financial ratios.
The availability of new data tools and the imperative to realize the value of data in an enterprise, will bring a new set of leading companies to the front – those that invest in making their businesses data-rich.
2. The coming shift from big data to big AI
Our traditional approach to building and using AI systems in the enterprise has been one that required a tremendous amount of “training” – a process that involves starting with a base-level AI engine and further modeling it with data from the enterprise to be able to deliver useful and useable recommendations specific to the context and problem at hand. But this can be a complex and costly undertaking, requiring significant investments in hardware, software, and personnel. And enterprises have struggled with large investments and long experiments only to get to accuracy levels below the threshold that can put it into mainstream lights-off production environments – and deliver on the promise of automation in the enterprise.
Big AI – built off Large Language Models (LLMs) trained on very large parameters is changing how enterprises think of and use AI. There are two reasons for this. First, the new models are so powerful out of the box that they significantly reduce the need for companies to build large data sets for and train their AI applications. This has significant implications for companies that previously had to invest significant time and resources in curating large data sets and training their own AI models. Second, generative AI is changing the way we think of and use AI – from doing machine-like tasks (for instance, machine prediction or pattern analysis) to human-like tasks (for instance, synthesizing or conversing), opening completely new areas for AI. With the availability of LLMs, companies can rely on these new models, eliminating the need to train their own. LLMs can continuously learn and improve as we use them, making them more accurate and capable over time.
I am already seeing companies throw away years of AI work because LLMs are already delivering better predictive performance out of the box. We will see more enterprises shift the center of gravity of AI work from modeling with big data to unpacking and using big AI.
3. Digital Transformation will be 80% People, 20% Tech
Article resource: https://www.forbes.com/sites/sanjaysrivastava/2023/02/10/data-and-ai-will-redefine-businesses-in-2023/?sh=100169fb4c17
Không có nhận xét nào:
Đăng nhận xét
Lưu ý: Chỉ thành viên của blog này mới được đăng nhận xét.