How Generative Ai Will Transform Businesses
John Titus Jungao, AI/ML Engineering and RnD Manager, Globe Group
John Titus Jungao, AI/ML Engineering and RnD Manager, Globe Group
Generative AI tools and products such as ChatGPT, MidJourney and Github Co-Pilot can potentially revolutionise how businesses work across various industries in the next few years. This technology can create text, images, audio and video that appear human-made and has produced impressive results. Furthermore, some of these outputs won art competitions and passed exams at leading law and business schools. As the technology matures and the adaption accelerates, its impact on business and society will only increase.
How do Generative AI-models Work
Generative AI differs from other AI systems by producing new data like images, text and audio rather than identifying patterns or making predictions. First, an encoder model transforms the source data into a lower-dimension representative array of numbers, called a token. Next, a decoder model recovers the original data using the token. A seed token (such as random numbers) is fed into the decoder model to generate new data. Next, another model measures how realistic the output is. As the model receives more data and computing resources, it improves continuously, making the results more realistic. Therefore, generative AI is an encoder-decoder system that improves iteratively to create more realistic outputs.
How Can Generative AI Impact Businesses
ChatGPT has reached 100 million users just two months after its launch and is used to answer questions ranging from food recipes and workout routines to scripts for building websites and Machine Learning models. Generative AI will impact businesses by (1) improving customer experience and (2) employee productivity and efficiency. Next-generation chatbots can incorporate customer experience as context and provide highly personalised recommendations.
Imagine a companion chatbot that suggests food menus based on personal preference and dietary restrictions or a bot that recommends and adjusts workout routines to account for recent injuries, medications and other physiological signals. As for productivity and efficiency, Generative AI tools can draft multiple ads and marketing material in a fraction of the time a team of creatives can do so manually. There are now existing generative AI tools that suggest entire lines of code to auto-complete functionalities. With this, employees work less on time-consuming tasks to arrive at the required product or solution faster. Overall, Generative AI has the potential to improve customer experience through personalisation and improve efficiency and productivity by reducing the time needed to build initial drafts.
Current Limitations
Before jumping right away into the bandwagon, adaptors of this technology need to be cautious. First, the output of generative models is “truthy” rather than truthful. Verification is required to check if the generated output makes sense. In generating images, digital artists still need to refine the results to remove artifacts (such as human hands having extra fingers). Similarly, text and code suggestions require a review from a human to ensure that the output makes sense. Therefore, while the draft creation speeds up, verification remains essential. Hence, the challenge for us humans is to develop subject matter expertise. Second, these models used datasets that may contain biases or prejudices upon training. Thus, models may unintentionally generate discriminatory and biased results. Third, the cost of fine-tuning these models to a company’s specific use case and scaling its use to the entire customer base can be huge. ChatGPT reportedly costs hundreds of thousands of dollars to run monthly. Hence, businesses need to consider cost-scaling when operationalising GenerativeAI applications/solutions.
It is important for companies to thoroughly evaluate the benefits and drawbacks of using generative AI tools in their operations and consider implementing appropriate measures to mitigate the risks
Conclusion
In conclusion, millions of users are already adapting generative AI tools for various reasons. As the technology improves, its impact will become more prevalent. Savvy companies can adapt these tools to transform their business processes in customer experience, employee productivity and efficiency and more. However, before going all-in with this new technology, companies must be aware of the limitations in output quality, potentially harmful biases and scaling costs. It is important for companies to thoroughly evaluate the benefits and drawbacks of using generative AI tools in their operations and consider implementing appropriate measures to mitigate the risks. Additionally, investing in the training and upskilling of employees to effectively use these tools can result in better outcomes and successful integration into existing workflows. As with any emerging technology, a thoughtful and strategic approach can help companies unlock the full potential of generative AI and stay ahead of the competition.
