The AI chatbot ChatGPT is a game-changer!

It’s only been a few months since the launch of AI chatbot ChatGPT by the firm OpenAI – but it’s already taken the world by storm.

Maersk technology is no stranger to new technologies. From Digital Twins using Internet of Things (IoT) for APMT, to artificial intelligence and machine learning (AI/ML) and robotics in Warehousing. All this is helping to digitise our physical assets and drive operational efficiencies, but ChatGPT takes AI intelligence to another level. So, what is it, and what are we doing about it?

ChatGPT is the latest in a series of AIs, referred to as GPTs, an acronym which stands for Generative Pre-Trained Transformer. This has a revolutionary AI language processing technology that provides a conversational interface that allows you to ask questions in natural language and get answers. It can follow instructions and has built-in safety features that attempts to filter out harmful content.

Holly added, "Significant leaps in AI at this speed are rare, and this breakthrough is a real cause to celebrate but it will create meaningful disruption for all, and we need to embrace and respond to the opportunity and the threat this poses to data security."

What’s the upside?

Maersk is investigating ways to use and invest in this technology to help improve lives for all - from dealing with the more ordinary tasks such as writing job descriptions, articles or educational memos, to providing customers with more self-help online.

How is Maersk using ChatGPT?

“We are currently using this generative AI technology on Maersk.com to auto-generate FAQs and improve accuracy of Search queries to increase traffic to the website. In the future, we see great potential for software code-generation or super intelligent ChatBots with language translation to help our customers get what they need online but also help our customer experience teams provide customers with the services / info they need, quickly and accurately. We have started to use this technology responsibly and where it makes sense to do so and where we can do this securely and within our policy to ensure the necessary safeguards are in place. As with any disruptive technology, we don’t know all the possibilities yet of how this will change our daily lives – working, education and otherwise. But we are investigating and taking a responsible leadership approach,” says Holly.

What’s the downside?

An advancing technology like this requires us to weigh up the excitement and benefits against any misuse and unfortunately, bad actors have emerged to create false information and identity theft by simulating individuals via phishing emails, creation of false websites - "Honeypots" and "Personas" while also spreading false stories and disinformation at unprecedented scale and speed. Therefore, we need to be cautious about using technology in the right way and guarding against misuse and take the necessary precautions to guard our customers and colleagues against this technology being used in the wrong way.

What’s our policy?

“We will continue to look at where we can adopt this amazing technology, while minimising the risk with the introduction of the Generative AI (ChatGPT) policy, that’s aligned with our Data Ethics,” explains Holly.

Generative AI (ChatGPT) Privacy and Cyber policy will ensure:

1. Data Security: By being clear what you can and can’t share with a ChatGPT. For example, sharing non-public (e.g., confidential, sensitive or any info that is not already public) is not allowed, nor is sharing Customer data or Intellectual Property (ideas, processes, technology, etc.). For example, do not share Customer service chats because it may contain pricing, supplier or contract information, or share legal contract drafts that may expose confidential business information.
2. Validation of content: “Human in the loop” validation is required. The ChatGPT responses must be validated to ensure the quality and accuracy of the generated content is guaranteed, before it is shared externally and being representative of Maersk.
3. Transparency to the User: ChatGPT can be perceived as a real person although not capable of functioning like one independently. Users should be informed that they are interacting with an AI language model and that the responses are generated by a machine rather than a human person to provide complete transparency that it is a chatbot that they are talking to.

What’s next?

The Generative AI (ChatGPT) policy has been circulated.

We are exploring an internal instance of ChatGPT and will publish a FAQ generated by ChatGPT in the coming weeks. If you have urgent questions, please email dataethics@maersk.com.