Three months after the introduction of ChatGPT, OpenAI has today announced the third major update to the AI platform: GPT-4. The rapid speed of development is a formidable problem for organisations: they want to move quicker than their competitors and include the newest AI tool, but they must do it responsibly—which is especially crucial when adopting potentially game-changing technology like AI.
In most talks, hundreds of company leaders indicate that they are wrestling with a business-critical question: How can GPT-4 and other analogous new technologies be optimally integrated?
GPT-4, according to OpenAI, is its “most advanced system, producing safer and more useful responses,” allowing users to analyse photos and mimic speech, and is intended to be the basic engine that drives chatbots and other systems. The business also stated that Microsoft’s Bing AI chatbot has been using the new software since its February debut.
So, what can businesses do to capitalise on GPT-4 and its successors? Regardless of the particular functionality of any one model, three major efforts will pay off:
Understanding The Underlying Technology
Understanding how generative AI works, its potential, and limits is the first step in effectively deploying it. The capacity of large language models (LLMs) employed in ChatGPT to produce text material on par with what a person might develop is its distinguishing feature. Its drawbacks have been well reported, and they include a lack of explainability—an LLM cannot specify its sources—as well as a proclivity for inaccuracy, both of which restrict their corporate applicability. This generation of LLMs has also been educated on unknown, widely generic data, which means they typically lack subject expertise required for a business application, such as defining pricing strategy in the healthcare market or enhancing bank productivity. And there’s always a healthy risk of producing inappropriate content, which we’ve seen several high-profile examples of in media and online.
Preparing for Governance
Crucially, now is the moment for businesses to develop and execute AI governance—a set of principles and processes that ensures an organisation strikes the correct balance between quickly implementing new technology and focusing on business objectives and avoiding risk. Businesses can evaluate relevant business applications, such as reducing IT overhead or speeding up data analysis, based on their potential benefit to the business, the resources necessary to build them, and any related risks.
The element of risk is critical to this study. Whilst LLM technology has helped software firms like Grammarly build great businesses, applying these technologies in a variety of corporations old and new is a whole new arena. Each company must choose how much risk it is prepared to assume in exchange for possible advantages and market leadership.
Infrastructure Preparation at Scale
Most people don’t realise how huge the most powerful LLM models are. GPT-3 and the recently published GPT-4 models are larger than standard machine learning models and may continue to expand exponentially. They are too costly to build and execute for all but the greatest tech businesses, and, in the case of the OpenAI models, are closed-source and only available via a paid API as a “model-as-a-service.” Building fundamental business capabilities on top of an API, as many who established their businesses on Facebook or other platforms have discovered the hard way, puts an organisation at the mercy of the API owner and is thus a significant risk for an organisation.
Because most organisations lack the resources to develop these models themselves, and because accessing a closed-source, pay-per-use model poses too much risk, many businesses will benefit from working with a smaller, open-source large language model, such as BERT, Flan, GPT-J, or other libraries provided by companies such as Hugging Face. Companies might create tremendous financial value by “fine-tuning” (i.e., modifying) these models on internal, specialised data, even if the platform can’t churn out award-winning sonnets on the side. Among the primary benefits might be:
1. Output that’s more specific and relevant to the organization. These models are particularly powerful in what’s called “few-shot learning,” meaning that the model only needs a few labeled examples to learn a domain.
2. More control over moderation to prevent unsavory or inappropriate outputs, while also improving the relevance of the response to the business.
3. All data stays within the organization’s firewall, helping meet confidentiality and data residency requirements.
4. Controlled costs for running the model, as the organization eliminates exposure to changes in API pricing from a for-profit supplier.
Although such a model lacks the extensive features of a general-purpose, huge language model like GPT-4, many of those characteristics are useless for focused corporate applications. Most service desks, for example, don’t need to imitate Hemingway’s voice or provide advice on Mexico holidays; they only require a concise summary of a larger transcript. While initial model setup necessitates specific skills, these models may later be deployed across the firm to serve practically all lines of business. Putting in place infrastructure to support such reuse is a wise prerequisite for the initial expenditure in establishing such a model.
Businesses have several infrastructure alternatives, ranging from in-house open-source models to the sole usage of models-as-a-service and everything in between. Smart techniques will allow firms to develop the correct strategy for themselves while yet giving flexibility to respond quickly when new technologies emerge and market conditions change.
Several businesses are drooling at the power of ChatGPT and looking for a way to use it to propel them to market leadership. To capitalise on this intoxicating potential, it will be necessary to build awareness, develop governance, and prepare a smart infrastructure.
The introduction of GPT-4 is a watershed moment in AI history, but in order for it to have a genuine influence on the workplace, enterprises must equip themselves to fully exploit its amazing possibilities.
Source: Forbes Technology Council
