What can ChatGPT do for your business?

Chat Generative Pretrained Transformer or ChatGPT, an AI chatbot and language generator introduced by OpenAI in November 2022, has been the center of attention since then. Amid all the buzz surrounding ChatGPT and conversational AI, numerous uncertainties remain about the true nature of generative AI and its potential impact on individuals and enterprises. This article aims to shed some light on these lingering questions.

What will ChatGPT change for businesses?

As a hyperautomation and AI innovation, ChatGPT is just one of many powerful tools that will shape the future of business and IT processes. It will be integrated into architected solutions that automate and augment human and machine tasks, providing powerful new capabilities for businesses to optimize their operations.

In addition to its potential for enhancing existing tasks, ChatGPT also has the potential to disrupt traditional job roles, redefining and even replacing certain tasks and activities that are better suited to automation or machine execution. As companies continue to embrace digital transformation and seek new ways to improve their efficiency and productivity, ChatGPT and other AI tools will undoubtedly play an increasingly important role in the enterprise.

What can ChatGPT do?

ChatGPT has many capabilities, including generating and improving prose and code, summarizing text, classifying content, answering questions, and translating language (including programming languages).

There are four main ways to deploy ChatGPT, but only two are currently available. The first method involves inputting prompts and receiving results through the web-based interface, currently the most commonly used approach.

The second method, called prompt engineering without APIs, involves using ChatGPT as part of a workflow and other technologies such as screen scrape and robotic process automation. This workflow can be created manually or with the help of other technologies.

The third method, prompt engineering using APIs, is not yet available but is expected to be released by mid-year. While some solutions may offer an API wrapper around ChatGPT, these are not recommended for production builds or scale, and OpenAI, the company that created ChatGPT, does not support them.

Finally, creating a custom build of the core GPT2/GPT3 or other foundation models for a bespoke implementation is possible. However, this would result in a loss of conversational interaction and prompt filtering provided by ChatGPT.

How will ChatGPT change the workforce?

It isn't easy to make a definitive prediction. Implementing hyperautomation and AI innovations, including ChatGPT, will lead to the creation of new job opportunities and the redefinition of existing roles. The overall impact on the workforce will depend on various factors such as the industry, geographic location, and the nature of the enterprise. However, utilizing such technologies will primarily target repetitive and high-volume tasks, focusing on improving efficiency, productivity, and quality control.

What are ChatGPT's current limitations?

There are several limitations to ChatGPT. Firstly, it was trained only on data up to 2021, so it may lack information on recent developments. Additionally, as it cannot cite its sources, its reliability depends on the sources it draws from, which may be inconsistent or incorrect. ChatGPT is also unable to generate images and lacks a supported API. It cannot be trained on personal knowledge bases and only makes predictions without understanding underlying concepts. Privacy concerns exist as it does not provide data privacy assurances. Despite recent improvements, it is not always reliable for mathematical tasks.

How secure is ChatGPT for enterprise use?

Gartner advises against sharing any personally identifiable company or client information that is not already available to the public on ChatGPT, as there are no guarantees of privacy or confidentiality. Employees using ChatGPT should treat any information shared as if it were being posted publicly. Additionally, it's important to note that any information shared on ChatGPT may be used to train the model further.

Rather than implementing a ban, we recommend creating a company policy regarding using ChatGPT. Many knowledge workers are likely already using it, and a ban may lead to employees using it secretly while providing a false sense of compliance to the organization. Instead, it's best to monitor usage and encourage responsible innovation while ensuring that the technology is only used to augment internal work with properly qualified data rather than being used with customers and partners in an unfiltered manner.

What’s next for generative AI?

In mid-January, Microsoft announced Azure OpenAI Services, including ChatGPT in the current quarter. This Azure version is expected to offer significant enterprise operational features, but the specific details, such as pricing and packaging, are still being revealed. This version is anticipated to incorporate APIs and provide greater privacy and security measures.

How should companies approach generative AI?

Companies should approach generative AI cautiously, recognizing that it is still in its early stages and that there is much hype around it. While the potential benefits are significant, companies should also explore other emerging generative AI use cases beyond language-focused ones.

It is important to encourage careful experimentation with generative AI, but not before defining usage guidelines, ensuring an understanding of the risks and issues, and having all generated text reviewed by humans. Companies should also create a task force reporting to the CIO and CEO to explore existential threats and major opportunities, plan a roadmap for discovery, and scope the skills, services, and investments needed.

In summary, proceed with generative AI but don't over-pivot, explore beyond language-focused use cases, encourage careful experimentation, and create a dedicated task force to manage the risks and opportunities.