Artificial Intelligence-Powered Large Language Models: Limitless Possibilities, But Proceed With Caution There are endless possibilities on the horizon, with the most sought-after goal being for AI to become sentient, and if this happens, humans will need to think of ways to control its power.
By Dhaval Desai •
Opinions expressed by Entrepreneur contributors are their own.
You're reading Entrepreneur Middle East, an international franchise of Entrepreneur Media.
When it was launched in November 2022, OpenAI's ChatGPT attracted one million users within the first week, and it has, at the time of writing, passed over 10 billion all-time page visits. As such, there's no doubt that as a tool powered by artificial intelligence (AI), ChatGPT is positively assisting companies across a myriad operations and departments, including but not limited to marketing, content creation, customer services, sales, human resources (HR), research, and development.
In the UAE, entities renowned for their digital transformation and smart services have been eager to embrace its application- Dubai Electricity and Water Authority (DEWA) is a key example here. AI technology is being used on its website through Rammas, DEWA's virtual employee, to enhance customer experience. But it's not just here- industry leaders from across the globe have been quick to champion its impactful contribution to their businesses. They are benefiting from increased operational efficiencies, stronger employee and customer relationships and more targeted and successful sales drives- indeed, some are even heralding it as a "lead nurturing superstar." Business leaders are also quick to point out how the provision of data, insight, and research materials, including competitor analysis, provided by ChatGPT, is even assisting in their strategic decision-making process.
But this is an approach that should be taken with caution. As with any new technology, especially where the impact is almost immediate, it's easy to be swept away by the initial utopian solutions it appears to be providing. But it's worth considering the fact that ChatGPT does have limitations that should be acknowledged and respected before its employment becomes the architecture on which the strategic development of a business and its future depends.
While ChatGPT is certainly a powerful large language model (LLM), it lacks the ability to reason or discern between factual and non-factual information, leading to occasional inaccuracies in its responses. Plus, in a world where real-time information is critical, with numerous platforms offering live news updates in both text and video formats, ChatGPT 3 -the basic version of the platform that's currently free to use- faces limitations due to its data being last updated in September 2021. This further underscores the importance of not blindly accepting information provided by ChatGPT without proper validation or further research from credible proven sources. It's also worth noting that ChatGPT is not particularly strong in mathematical tasks either, an area where its competitor, Bard by Google, outperforms it significantly. This was also a contributory factor to some major US banks, including JP Morgan, Goldman Sachs, and Chase of America, heavily restricting its use by employees.
Another consideration for businesses, as the fast-paced evolution of this tool and its capabilities gathers pace, is the ability of information technology (IT) departments and individuals to keep abreast of it. The launch of a new generation ChatGPT -which was called AutoGPT- in March this year, introduced AI that has the ability to reach goals through its own reasoning, putting it in a similar calibre to what humans do, and how they process information. For instance, AutoGPT can delegate tasks to various other services. An example would be booking a direct non-stop flight from Dubai to Toronto where AutoGPT will delegate tasks to various other services such as Kayak, and then be fully connected with their APIs to make a booking on behalf of the user. Again, for businesses, the benefits of AutoGPT are limitless. Though accessibility to the tool is not currently as straightforward as ChatGPT (which is available simply via a browser), there are tools like AgentGPT that allow AutoGPT to be accessed via a browser.
In the meantime, the Transformer architecture, which underpins modern LLMs, is also becoming increasingly popular and accessible. A notable example of this is the very recent release of the iPhone 15, which utilizes Transformer architecture in its neural engines to perform complex AI tasks directly on the device, eliminating the need for an internet connection. A future where neural engines advance to the point where all processing occurs locally, without the necessity of internet connectivity, is thus eagerly beckoning. At the same time, there are endless possibilities on the horizon, with the most sought-after goal being for AI to become sentient. If this happens, humans will need to think of ways to control its power. It will then become the responsibility of business leaders and their employees to skill up and utilize this power to their advantage, or risk becoming obsolete.
On an encouraging note though, a recent study carried out by Statista revealed that, globally, individuals under the age of 34 account for 60% of ChatGPT users, with those under 24 the second largest userbase. As they represent the next generation of business leaders, it's reassuring to know that their familiarity with -and use of- the emerging technology is so prevalent, which augurs well for the future.
Related: Reality Check: The Current State Of Play For The Metaverse
