Generative AI: What it Means for the Industry

OpenAI's latest offering of ChatGPT (built on Generative Pre-trained Transformer using a deep neural network) has triggered a spate of new research and development. And debate.

Even as tech companies scramble to build smarter AI tools - especially chatbots - governments, educational institutes, and many corporates are facing the new age Shakespearean dilemma: to use or not to use ChatGPT, and how much to use.

Undoubtedly, the chatbot in its latest young avatar is amazingly useful: composing smart emails, scheduling meetings to being virtual shopping assistants, and even in planning your holiday trip!

We do have companies use chatbots when short-staffed. Or help walk-in customers at a front office when a professional is not present. However, recently, concerns were raised over algorithms used in AI-driven predictive analytics tools by a US state government welfare agency. The risk scores flagged by the predictive model were found baseless and unreliable in some investigation cases.

Some top-notch financial companies are planning to use ChatGPT for a personal touch for answering frequently asked questions and for an enhanced consumer experience. Many large financial companies are integrating it in their customer management systems. AI-driven predictive analytical tools have found extensive use in risk management and decision-making. However, even today, existing chatbots are often seen to miss the context, cannot detect humor or sarcasm in the prompts, and often give a different answer.

AI-driven tools help doctors and medical professionals in clinical diagnosis and planning, and choosing a treatment plan. As in bio-medical research, though the results have high accuracy, these often contain outliers and require further validation by experts.

Even in these established systems, further improvement in algorithms is required before these can be integrated into other systems or become fully reliable. So with better software development, AI-driven tools and systems can play a better role.

Perhaps a good fit currently seems to be in the education industry. Some institutes are using ChatGPT for its virtual and classroom assistants. In fact, students have taken it as duck to water. Using it for writing essays and papers. Though, as in one case, a group of students in a US university working on a given subject submitted similar papers! In fact, some universities have banned the use of AI-driven tools on campus and their use by students.

Many educational experts are concerned that frequent use of AI tools may affect learning abilities.

For some time now, AI-driven tools are being used in career counselling. It helps many students get a direction. However, many students are not convinced with the results. A few who chose streams based on the results have found to have quit those later!

Bloggers and content writers too are taking help from ChatGPT. But, many bloggers are wary of using it too much as 'it lacks the quintessential individual touch'. Bloggers swear it standardizes writing. Though it gets better with repeated prompts. Though it does help writers start off, get a new perspective, and strengthen their writing.

We can say those who are using it smartly are benefitting from it. Even improving their productivity and quality of work. For example, when facing a writer's block or struggling with an introduction or conclusion. Writers can use the content rolled out by ChatGPT as a pointer or for baselining their writing. Thereby, not losing their hold on their work.

Tech writers are wary of even attempting to use it for introduction or structuring content. Given the highly company-specific information they deal with. Confidentiality of data is of prime concern. Information security cannot be compromised to any degree. For certain industry terms and definitions, they still prefer to Google it to know the correct meaning and write these in their own simple terms for their audience.

What We Think:

AI in general and ChatGPT (current versions) specifically is not a one size fits all. Decision-making and risk management decisions should not be done on AI-driven tools alone. ChatGPT may help an individual not so well-versed with writing to draft that important email. However, it lacks real-world knowledge. Its information may be outdated, outright incorrect, often irrelevant, and at times incomprehensive as ChatGPT is based only on data it is trained on. As data security is not guaranteed, it needs to be used sparingly for company information. Also, its complete integration with other systems is yet to be ascertained. It will evolve into something bigger and better. But still, these are early days.

Watch this space for our next insight into the most current topics in the tech and content world.