Suits The C-Suite

(First of two parts)

In November, ChatGPT 3.5 was released — and took the world by storm. ChatGPT touted its ability to create essays, write computer code, pass board exams, create business plans, and many other tasks such as, but not limited to, analyzing professional contracts and complex spreadsheets.

The ChatGPT 3.5 architecture follows ChatGPT 3, which was launched in 2020 and is now being used by many large organizations such as Microsoft Corp., Google LLC, and, Inc. for chatbots, virtual assistants and other AI-powered applications.

ChatGPT, or the Chat-Generative Pre-Trained Transformer algorithm, was developed by OpenAI. OpenAI is an artificial intelligence (AI) research laboratory based in the US established in late 2015. It aims to promote and develop friendly AI, running on the fifth most powerful supercomputer in the world.

Close to six months after the launch of ChatGPT 3.5, OpenAI also launched ChatGPT 4. It promises to be even more powerful and versatile than its predecessor, improving on the weaknesses and limitations of ChatGPT 3 (which uses a relatively small database to train on). ChatGPT 4 uses a much larger 50 terabytes of high-quality training data through a combination of automatic and manual curation methods. This allows ChatGPT 4 to deliver better conversational AI applications, understand context, and generate more natural-sounding text. It is powerful enough to detect and respond to changes in tone and sentiment, and unlike ChatGPT 3, can also make images.

Generative Pre-Trained Transformer (GPT) is a type of large language model (LLM) neural network that can perform various and complex natural language processing tasks. It is a type of a deep learning algorithm that uses a transformer network (a sequence to sequence translator architecture used for language models and computer vision), specifically developed to train from large quantities of unlabeled text using unsupervised learning, analyzing patterns in the data set to generate human-like text in response to input.

LLMs need access to large datasets of text called training data. Such data come from a variety of sources including books, articles, websites, academic papers, social media posts, blogs, news articles, and other online and offline text sources — without any explicit supervision or guidance on what to learn, except to automatically discover patterns and relationships in the data and use them. ChatGPT uses this data to generate more natural-sounding text.

While there are many positive opportunities presented by ChatGPT, ongoing debates in the tech community center on the threats posed by the larger AI. ChatGPT is indeed revolutionary, but it also gave us a taste of the real risks and dangers.

Some of these hot topics relating to the human aspect include social manipulation, job losses, social surveillance, gender and race biases, socio-economic inequality, weakening ethics and goodwill, financial crises, and a dangerous arms race of AI-powered weaponry.

ChatGPT responses can be categorized into those that are mathematically or scientifically accurate, i.e., the answer to 1 + 1, or that water is liquid at room temperature. The other category consists of subjective responses, i.e., whether red is a better color than maroon, or whether certain politicians are performing better than their predecessors.

It is worth noting that there have been concerns about the potential biases in the training data sets used for language models like ChatGPT. Biases in the data can lead to biased outputs, which could have negative consequences in real-world applications. ChatGPT, just like humans, can still provide subjective, inaccurate, or wrong answers that are biased. When these biases cross ethical boundaries because of the quality and manual curation of the training data, this means that such biases can sometimes cause more societal harm than good.

Rest assured, ChatGPT (and AI) will be here to stay, continuing to evolve and advance at lightspeed. It will continue to highlight that the world we live in will be significantly different as early as next year. Many businesses are scrambling to understand both the implications and opportunities provided by ChatGPT to their organizations.

ChatGPT as applied in business could, in a lot of ways, improve the bottom line, enhance efficiency, and transform customer experience while reducing costs. Some use case examples for ChatGPT are chatbots, content creation, code development, fraud and abnormality detection, language translation, voice assistants, and hyper-personalization for recommendation engines. There are also potentially vast opportunities, along with accompanying risks, in sectors such as education, creative services, professional services, content creation, and many others.

Many more technically adept companies are already finding amazing use cases of ChatGPT and AI that end up disrupting traditional businesses.

There is no doubt that ChatGPT is still in its infancy stage, which simply means that there is much more to expect. Our lives will change, and the rapid rate of this change will be like no other compared to all human history. Just like electricity and water, ChatGPT is also expected to become a mainstream utility. It will be much faster, cheaper, more accurate, and eventually, some even say it will be sentient. It will become a necessary and unavoidable part of our daily lives.

According to a report from Opus Research, 35% of consumers would like to see more enterprises incorporate AI tools like chatbots, whereas 48% of them are indifferent as to whether an AI or a human were to assist them. While not the majority, a considerable percentage of people are seeing the benefits of AI. As this technology only continues to get better, many jobs and traditional businesses will need to transform or be at risk of being displaced. Industries and processes will be disrupted, and new opportunities and applications will surface. The only question will be: are we ready for it?

In the second part of this article, we discuss the practical ways ChatGPT can be used in business and the potential risks it presents.

This article is for general information only and is not a substitute for professional advice where the facts and circumstances warrant. The views and opinions expressed above are those of the author and do not necessarily represent the views of SGV & Co.


Randall C. Antonio is a technology consulting partner of SGV & Co.