ChatGPT: CIOs are yet to figure out its enterprise usage

We've been talking about how artificial intelligence (AI) might change the world in such a way that jobs like car drivers, personal assistants, customer service representatives, consultants, instructors, and even authors like myself might become extinct. These possibilities abound in science fiction films. In practicality, as 2023 got underway, people saw a number of these opportunities.

Everyone is ecstatic about a new AI bot called ChatGPT. Even those who haven't extensively used a virtual AI assistant in the past or aren't too familiar with AI as a technology have a reason to pay attention to this powerful Generative Pre-Trained Transformer (GPT). (See: ChatGPT Reignites Web-Search Wars)

Just a few months after its debut, the ChatGPT has already been used by more than 100 million users, according to several media reports. Its launch by many has been seen as a competition to the dominance of the undisputed web search giant, Google. Microsoft probably has seen this as an opportunity to up the ante against Google by announcing an enormous investment in OpenAI, the maker of ChatGPT, already. Reports suggest that Bing, Microsoft's web search engine, will soon run on ChatGPT, which makes the scenario much more intriguing.

But the focus of this month's cover feature is not on ChatGPT's impact on Google's hegemony. Given the excitement surrounding ChatGPT, organizations must determine whether or not ChatGPT can be successfully integrated into their IT environment to create cutting-edge enterprise apps. Would it be the new game-changer and play a significant part in improving organizational procedures as well as their social media/customer engagement strategy, e-commerce goals, and customer service?

What’s so unique about ChatGPT?

The ChatGPT can respond to a wide range of queries and can even create an in-depth essay on just about any subject you can think of. The technology leverages the vast amounts of data it was trained on to identify certain patterns and make intelligent judgments. This data is based on online data and through well-certified data sets.

Most chatbots that people deal with still lack understanding and can only answer simple queries or provide minimal assistance to users with their problems. But, according to industry experts, the field of natural language processing seems to be entering a new stage as Chat GPT shows the technology's capacity to sustain a discussion through multiple questions in addition to producing software code. The AI tool can let users easily write detailed essays, provide grading/feedback on their work and even solve some mathematical problems.

“It is important to note that GPT has the potential to greatly improve the efficiency and effectiveness of health care by providing personalized recommendations and treatment plans to patients and helping healthcare professionals with their daily tasks. While there are challenges to overcome, such as the need for accurate and up-to-date data and concerns about privacy and security, the use of Chat GPT in healthcare holds great promise for the future,” says Gaurav Bagga, SVP-Product & Engineering at Pristyn Care.

Gaurav sees a mix of risks and rewards of the intelligent AI interface by adding that some possible future applications of GPT in healthcare could include using GPT-powered chatbots to provide mental health support to patients or using GPT to generate personalized exercise and nutrition plans for patients based on their individual needs and goals. “GPT could also be used to help healthcare providers with tasks such as scheduling appointments or managing patient records, making the healthcare system more efficient and effective.”

ChatGPT and similar tools can be trained on organizational data to transform the sector as technology advances. They also serve as good beginning points for software development, content creation, knowledge management, customer interactions, and strengthening employee experience. Virtual personal assistants, tutors, and event planners are all concepts that could materialize.

It's still a work in progress...

Even while ChatGPT can greatly boost human productivity with the correct use cases and implementation, it still has several user-facing issues that keep the majority of tech titans on the fence about using it.

"We developed a Chatbot in 2018 and are still using it for our retail customers. We have made significant improvements to it. What began as a simple virtual assistant has developed to take advantage of AI's intuitive potential to construct algorithms that could be moving toward deep learning to provide a more personalized experience. We are still exploring how ChatGPT might change the ecosystem and how it can enable us to help our consumers and business more effectively," says, Yatish Shivaprasad, Director & Head - Corporate Functions Technology, at Societe Generale Global Solution Center.

CIOs and IT leaders are mostly playing a waiting game to assess ChatGPT's potential, even though the tool has attracted the attention of the tech elite and has the potential to change the game in the future.

Indian Oil has made substantial use of the bots to increase utility and efficiency. The country’s flagship national oil company is looking at the developments with a lot of interest around ChatGPT but noted it's too early to comment on a specific use case. "We are looking into taking on AI in a big form and these kinds of things [ChatGPT] where a lot of interactions happen in the business side - with your customers or internally. We are closely monitoring the evolution of ChatGPT, which is an intriguing development,” says Manish Grover, Executive Director (Strategic IS &IS), IOCL.

See what ChatGPT has to say about its applicability to business use cases.

Business-to-consumer (B2C) use cases, such as developing chatbots for customer support, product suggestions, and marketing, are likewise seen by industry experts as being better suited for ChatGPT's current use.

"There is more value for ChatGPT on the B2C side. From a B2B kind of business, that may not be a great potential. That could not have much promise from a B2B type of business. We aren't actively working on it right now, and we don't have any urgent plans either [on ChatGPT]. However, we are developing our data lake and expanding it by including data from numerous additional sources. Additionally, we are implementing analytics and constructing AI mechanisms to leverage this data and develop quality insights," says Amit Pradhan, Vice President- IT and CIO, Dixon Technologies, a leading Indian multinational electronics manufacturing services company.

Key challenges

Despite all the fanfare around the ChatGPT, it still has a ways to go before it can be considered a game-changer for enterprise usage. Chat GPT's reliance on a sizable text training dataset is one of its primary drawbacks.

"I believe it's too early to see ChatGPT as a strong business case. However, as the popularity grows up in the future then certainly we have to take evaluate its feasibility in the Indian banking ecosystem and we have to brace up for the challenges associated with it. For this, the entire financial ecosystem has to be on a single platform. For instance, NBFCs, banks, fintech, account aggregators, and data aggregators all need to be connected,” says Hemant Verma, CGM IT and CTO, Punjab National Bank.

Agrees, Jitendra Mishra, CIO of Akums Drugs and Pharmaceuticals “You will see a big adoption in the future. But it's a very early stage to comment. Having said that, ChatGPT has a great sense of integrating most of the processes. In our environment as well, we are looking at that kind of technology that can harness the power of AI effectively and help us deploy intuitive applications but it's still way too early.”

Some of the key challenges that the Open AI platform needs to address include:

Inability to distinguish between culture and national boundaries: The patterns and structures of natural language can be understood by Chat GPT thanks to this dataset, although it may not always accurately reflect the diversity of languages and cultural experiences seen in the real world. According to Gaurav Bagga, due to this limitation, Chat GPT may struggle to understand and respond properly to inputs that are significantly different from the data it has been trained on. This can lead to inconsistencies in the model’s performance and may result in responses that are inappropriate or unrelated to the user’s input.

Users are unable to access real-time information since Open AI cannot connect to the internet, and even ChatGPT is unable to access location-based information. To create killer apps for businesses that can respond depending on the most recent data, it must be coupled with other systems and APIs that can provide it access to real-time information, such as a database or cloud service.

Misleading or inaccurate information: Another key concern for CIOs is that ChatGPT largely relies on machine learning methods, hence the quality of its answers depends on the data it was trained on. The tool may produce replies that are inaccurate and misleading if the training data is biased or flawed. It is also not possible to completely rely on ChatGPT for important conversations and believe that this will completely automate any process or communication. There has to be some form of monitoring by a human since the content produced by ChatGPT is not completely authentic. Some of the information may be based on outdated or inaccurate data, or it might not contain the most recent or confirmed information.

For instance, ChatGPT provided an erroneous response when we typed a query about a well-known cricket player below.

Well, the answer was incorrect. The intelligent conversational tool was prompted once more.

Let’s take another example.

For ChatGPT, we asked the following question with a straightforward computation, and he responded properly on the first try. However, the tool apologized and accepted the defective version we sent after we purposefully mislead it into believing it had sent the wrong response.

It can be quite problematic to use these automated technologies in their current state without human involvement. Imagine this AI-powered tool recommending the incorrect medication to a patient at odd hours or providing incorrect information to a student studying for a competitive exam. To further improve the quality of its responses, careful integration of APIs and particular extra tools is required.

Managing security aspect: In the already complex security ecosystem, managing the security of conversations is the other big challenge. Users frequently divulge personal information in a conversational setting, such as their educational background, current health, or preferred brands of goods or services. As such, technology leaders believe that they need to think through to take extra steps while creating a chatbot based on ChatGPT to ensure that the personal information of their users is protected. Knowing OpenAI's privacy standards is essential, and hence the platform's potential use of users' data must be carefully considered.

The technologies may also contribute to the spread of cyberattacks, but that is a subject of a different debate. A study by BlackBerry found that even though respondents in all countries believe ChatGPT is generally used for "good" purposes, 74% of them recognize its potential cybersecurity threat and are worried about it. BlackBerry is a company that provides intelligent security software and services to businesses and governments around the world. The ability of ChatGPT to assist hackers in creating more convincing and authentic-sounding phishing emails is the top global concern (53%), followed by its ability to help less experienced hackers advance their technical knowledge and develop more specialized skills (49%) and its use for disinformation (49%) dissemination, according to the study.

Large OpenAI: The new OpenAI platform is also large and needs a lot of resources to run in addition to the above challenges. Industry experts have cautioned that because of this restriction, using it in some applications or on some devices is exceedingly difficult or unfeasible, especially in environments with limited resources such as low-memory mobile phones.

Ready for a brave new world albeit with caution

Without a doubt, tools like ChatGPT will have a huge impact on the future, but for businesses, relying only on them might be risky. Businesses may use AI-powered chatbots in the future to streamline staff onboarding and offboarding, reinforce recruitment procedures, and assist with consumer complaints, particularly during odd hours.

According to Deepak Kumar, Founder Analyst at BMNxt, the Chatbot is far from being mature currently but is likely to be evolved into a credible prototype of what may be termed an interactive search and query engine. “The ultimate maturity and success will be a function of not just the AI algorithm lying underneath. It is therefore a domain Google is eyeing quite keenly,” he opines.

However, it will be disastrous if someone believes that technology will completely replace the human mind in certain areas. “The human element is still necessary for reviewing outputs and developing messages that are aligned with a firm's business plan and vision,” Gaurav Bagga states.

It would be important for CIOs and technology leaders to stay up to date on ChatGPT-related technological advancements, but it would also be a huge task for them to ensure that their staff members continue to receive training and exhibit innovation in how to use these technologies safely for organisational efficiencies rather than becoming complacent and relying solely on them.

- The author can be reached out via email at Jatinder.singh@9dot9.in