By Shailesh Dhuri
Generative Pre-trained Transformers or GPTs as we commonly know them, are a powerful family of neural network models, leading the way in generative artificial intelligence. GPTs are already having a fundamental impact across many walks of life – the way we work, the way we think, among others. While scientists have been attempting to develop generative tools in the Artificial Intelligence space for decades, the recent success of GPTs can be attributed to three critical factors:
– Phenomenal Computing Power: The exponential growth in computing power from the advancement in technology infrastructure over the last few years has enabled speedy matrix multiplication, a core element of machine learning, at a low cost with high output.
– Global Internet Expansion: The widespread penetration of the Internet, especially in developing nations such as India, has resulted in the generation of enormous amounts of data, enriching the capabilities of AI applications through large datasets.
– Advancement of RNN and ANN: Recurrent Neural Networks (RNNs) and Artificial Neural Networks (ANNs), which have been in existence for 15 to 20 years as classifiers, can now perform advanced predictive tasks due to the availability of extensive training datasets.
Integration of GPTs will inspire various industries and organizations to formulate strategies. At present, organizations have access to large amounts of accumulated knowledge through internal and external data. This data is a valuable mine of training sets that GPTs can use to build intelligence and predict future business outcomes. With GPTs, analysts and users can now ask questions to the entire corpus of the knowledge of an organization in real time and interpret the results to help the management in strategy building.
GPTs have Positively Impacted the Analytics Landscape
GPTs are making a significant impact across various industries, leveraging spoken human languages and structured logical thinking to generate predictive responses. This has resulted in GPTs solving problems that were not addressed by previous algorithms. One of the biggest differences between earlier prediction models and GPT-driven models is that GPT understands context and is able to apply logical reasoning before making predictions. Some real-time applications of GPTs include:
– Interactive Chatbots: GPT-powered chatbots have transcended the limitations of menu-driven chatbot systems. GPT-powered chatbots are able to understand the context before interacting and execute high-performance tasks. Interactive chatbots will have a positive impact on many businesses resulting in improved employee and customer satisfaction and reduced costs.
– Scientific Research: GPTs have opened-up significant opportunities in scientific research. With GPTs in action, creating vaccinations for viruses will become transformative under known adverse reactions using Monte Carlo simulations and based on individual DNA sequences.
– Revolutionize Communication: Integrating 5G with Large Language Models will help in optimizing communication networks by understanding language signals and making predictions. This will reduce the cost of mobile telephones and the Internet significantly.
– Radar Technology: GPTs have the potential to develop “only listening-mode radars,” improving information retrieval and predicting outcomes without disclosing radar locations.
GPTs are increasingly disrupting traditional analytics and enabling organizations to get advanced insights, make logically sound predictions and gain a competitive advantage. However, consuming data and predicting outcomes require infrastructure and systems with high computational intensity, and as such GPTs need to be integrated with high-powered and scalable computational engines. While this has not been a deterrent thus far, the pace at which the GPT space is advancing, efforts will have to be made to continue developing the supporting infrastructure and technology.
GPTs will Elevate Job Profiles, not Propagate Job Losses
Before delving deeper into this apparent challenge, we must first understand the concept of ‘Job to be done’ and ‘Task to be performed’. The job to be done is the outcome that we want to achieve while the task to be performed is the task we need to accomplish to achieve that outcome. For example, if commuting from point A to point B is the job that needs to be done then the means to reach from A to B is the task. The means of commuting have evolved over time from walking to bicycles to motorcars to metro trains and so on. This means that in this entire journey of transformation and evolution, we need to constantly upgrade ourselves to be relevant. GPTs are all about this next phase of evolution.
In the analytics space, the role of a data analyst is to provide relevant insights, previously it was achieved using tools such as Excel and dashboards, among others. In the world of GPTs, mundane tasks will be automated, thereby empowering data analysts to focus on providing valuable and new insights from the data at hand. GPTs will save massive amounts of data analysts’ time which currently is spent on preparing data for analysis. In a nutshell, GPTs will help elevate the job profiles of many of the roles that we see today, provided we are ready to embrace this change to evolve and upgrade.
GPTs are not the Future, they are the Present
In conclusion, GPTs represent a revolutionary force in analytics, empowering organizations to gain a competitive advantage and drive efficiency. While they promise to revolutionize computations on a massive scale, they will not replace human intelligence but rather complement it, ushering in a new era of augmented analytics. The future holds great promise as GPTs continue to evolve, transforming industries and human potential alike.
(Decimal Point Analytics is a leading management consultant in financial solutions to clients worldwide. It provides Research, Data Management Solutions & Advanced Analytics Services, based on Artificial Intelligence, Machine Learning, and Automation. The authore is CEO Shailesh Dhuri is an MBA from IIM Bangalore, and the views expressed in this article are his own)
