ChatGPT, what businesses should be considering now.

 

The release of the generative AI model, ChatGPT has many considering what the future of their job, and their business, might look like.

Late last year, AI technology took a huge step forward. In the first five days of OpenAI’s release of ChatGPT in November, an AI chat bot that generates answers to almost any question via user input, more than a million users signed up. In comparison, it took Instagram two and half months to reach the same numbers. The success of the model has even caught its creators short, with the error message ‘ChatGPT is at capacity right now, please check in later’ often appearing. But what’s so different about ChatGPT from other AI models that has caused its popularity to soar? 

This chat bot is considered much more advanced than previous iterations because of its use of natural language processing rather than rules-based answers. This means that essentially, it’s a computer that can answer any question in a way that humans can understand and a tool that anyone can use - no computer science degree required.

When first released, people on social media had fun with the program. It was asked to write instructions for how to remove a peanut butter sandwich from a VCR in the style of the King James Bible, write poems about Disney movies in Latin and explain Karl Marx’s theory of surplus in the style of Taylor Swift, all of which it did with unnerving ease. Its real-life and commercial applications are only just being explored and understood. Ryan Reynolds asked it to write a commercial for his mobile network company in the style of Ryan Reynolds, which he dubbed ‘mildly terrifying’, universities in Australia are actively considering how they can change their assessments after students used it to write convincing exam papers and coders are asking it to check for mistakes or to in fact write code from scratch.

 

The advent of this form of generative AI which learns from vast amounts of unstructured data and can produce ‘original’ content in the form of copy or images (DALL.E is an image producing AI system also from OpenAI that can generate images of almost anything you can imagine). The experience has an uncanny valley vibe about it and has many artists and writers contemplating their future careers.

 

There is no putting this genie back into its bottle, so business leaders need to consider how they can best utilise the technology to enhance their operations or consider how the current business models may have to change to take advantage of this leap forward. How we adapt to best use this technology could reshape businesses and roles.

The most obvious place to begin is to consider where AI is already being used within the organisation and consider moving to a generative AI model.

 

When we asked ChatGPT how it could be used to improve business operations it offered:

 

Bettering customer service is an area that businesses are already leveraging AI to assist through the use of chatbots. A GPT-driven chatbot can provide better results for answering common questions.

 

Generative AI chatbots can also be used to help with lead generation. They can engage with potential customers to compile their contact information before being assessed of their individual needs.

 

The model can also be used to personalise information for customers to nurture leads and build customer relationships and finally, the analytical capabilities of the program can offer data and insights on customer interaction, needs and pain points that can help organisations improve products, services and processes.

 

However, there was a caveat. It finished with:
 

It’s important to note, however, that GPT models need to be fine-tuned to the specific domain of the application for best results. Additionally implementing the chatbot should be done with an understanding of the limitations of the model and how to handle situations where the model generates inaccurate or irrelevant responses.

As with all new technology leaps, ChatGPT is not without its red flags. Leaving ethical and intellectual property arguments aside, one of the current biggest issues is its ability to serve up incorrect information, with apparent confidence. Misinformation, and the spread of it, is a growing challenge for our communities, indeed our democracies, and ChatGPT has the ability to not only offer, unintentionally, the wrong information. Those who are intentionally looking to spread incorrect information can do so on a very large scale, for very little cost.

 

Additionally, the tool has only be trained on data up until the end of 2021, so it has no knowledge events or break throughs that happened in 2022/23.

 

Businesses looking to implement ChatGPT, or any AI models into their operations need to consider the inherent risks involved – and the task of identifying those risks goes way beyond just the IT team. It needs to involve legal, risk, security, analytics and even manager across the entire business who can monitor vigilance at the front lines. AI is in fact of whole-of-business concern.

Will the machines be taking over our jobs? No, not right now. These programs can mimic humanness, but they can’t understand. It needs a person to prompt it to do anything, it needs humans to fact check and rewrite nonsensical sentences – and it’s going to need humans to ensure the right guardrails are in place to stop unethical behaviours.

 

But it’s ability to kickstart and amplify the work that is already being done is a potential benefit that’s hard to ignore, especially at a time when staff shortages are still being acutely felt in many industries.

 

Many are already reporting that they are using ChatGPT as the starting point for their work, ie to write a catchy subject line for an email, or to do topic research around key words for SEO, however the work of the human starts next. There are some predicting that there is going to be a sudden onslaught of huge amounts of not-very-well written articles, which offers brands an opportunity to stand out and gain trust from their readers.

 

For organisations looking to take advantage, start with a strategy document that considers the benefits, risks and opportunities and then building a deployment roadmap for AI foundation models. In the first instance, this will help identify if the benefits outweigh the risks for the use cases you’re considering.

 

Success depends on identifying the application of this technology in a way that offers meaningful benefits to an organisation.