Last November, the virtual chatbot’s prototype was launched and its rise has been meteoric. Experts have reacted with concern or lauded its capabilities, sometimes in the same sentence. 

The chatbot responds to questions or requests at a dizzying speed, extracting answers from the vast swathes of information on the internet. It can write documents by request, analyse text, translate languages or summarise long and detailed passages.

It can also be trained to code software, screen information, generate personalised messages, help with administrative tasks, provide feedback, identify biases and much more.

“There are so many good AI options. ChatGPT is just the one that has taken the world by storm,” says Baker Tilly Staples Rodway National Technical Manager, Audit & Assurance Services, René Lamprecht.

One thing is clear: AI is here to stay and business that don’t embrace it risk falling behind. And while many are apprehensive about the threat it poses to their employment or business, it’s easy to forget that the biggest immediate threat might be the people and companies that quickly become skilled at using it, she says.

However there are potential downsides to the likes of ChatGPT, so it pays to understand and mitigate them.

Things to be mindful of with using ChatGPT in your business

One of the immediate and predominant issues is patchy accuracy. Most of ChatGPT’s errors can easily be corrected when you check or test results, but some could have serious repercussions, such as its recent claim that the whistleblower in a high-profile Australian bribery scandal was the perpetrator, potentially leading to a lawsuit against its developer, OpenAI. Other existing or potential concerns include:

• ChatGPT can collect and process vast amounts of data, raising misgivings about privacy and security if sensitive information is accessed and used by third parties, or used to train AI. This can be mitigated by creation of policies around permissible use within your business.
• Systems should be designed to protect personal data, and it pays to be transparent about the use of any data collected, including clear opt-out instructions for clients.
• If AI is trained on biased data or designed with biased algorithms it can generate unfair outcomes, such as discriminatory hiring or loan approvals. For example, a recruiting system might associate certain qualities, skills and experiences with specific demographics if the training data is not sufficiently diverse and representative of the population.
• As artificial intelligence becomes more autonomous, it can be difficult to determine who is responsible for AI-based outcomes. To address this, systems should be designed with transparency in mind, and be subject to ethical and legal oversight.

With those cautions in mind, there are also the obvious benefits of ChatGPT, with its clear, rapid, detailed and often quite accurate content. It’s staggering even in its infancy.

Implementing ChatGPT into your business

Baker Tilly Staples Rodway Taranaki special services director Rob McEwan says some key things to remember when using ChatGPT are to provide very specific requests for optimal results and when using the free plan, don’t ask for information beyond its September 2021 knowledge cut-off. Also consider that it might not remain freely available for commercial purposes. “ChatGPT isn’t the answer to everything, but there are opportunities,” he says.

The best way to implement it – or AI in general – is to explore its potential by testing it in a controlled environment. You can then see where it provides the most value and gradually increase your AI capabilities as technology advances.

Pre-trained models can be fine-tuned for specific use (reducing costs and the need to train from scratch) or trained to meet many of your specific business needs. However, the latter requires technical expertise and can be quite expensive.

It’s important to start with a clear understanding of your business goals and how the technology can help. Look for a platform that aligns with your goals, is user-friendly and offers customisation, and investigate costs. Then you can train it using historical customer interactions, product descriptions, customer reviews or other data sources.

Lastly, integrate it into your business processes then regularly monitor its performance for optimisation and to identify any biases, inaccuracies or technical issues.

Will AI take my job?

Jobs in some industries are clearly more at threat, but there are many variables, including AI’s rate of development and the breadth of tasks it will be able to perform across any one role. Bear in mind that it can enhance work capabilities and may also generate new jobs. Meanwhile, there are still things humans can do better, such as:

• Understanding and expressing emotions, empathy, and compassion, and interpreting and responding to nuances in communication.
• Understanding when data is sensitive and should not go on the record.
• Driving creative processes with imagination and innovation.
• Applying common sense, intuition and judgment to make decisions and solve problems.
• Carrying out tasks that require physical dexterity and versatility.

Adopting new technology always presents challenges, but stagnation is risky and businesses that don’t stay abreast of artificial intelligence will be left behind.

AI is transforming the way we work and live. While it’s understandable to feel apprehensive, if we stay informed, use it wisely and seek professional guidance where needed, it should be an ally and a tool that unlocks endless opportunities, growth, innovation and efficiencies.

Click below for Baker Tilly International’s recent look at AI – including the way it could disrupt digital giants like Google.