The AI revolution is coming at us like a freight train, reshaping industries and redefining the way we approach business operations. This shift presents a myriad of opportunities and challenges that businesses must navigate to stay competitive in a rapidly changing landscape.
Generative AI technologies, like Chat GPT, are allowing companies to streamline their operational processes and focus on leveraging unique skill sets. Things like expense tracking, data analysis, customer support, and appointment scheduling, give AI the potential to add great value to every business across the country.
One individual at the forefront of this revolution is Dr Paul Duignan, a Psychologist and AI Strategist who is charting new territories in the field and shaping its trajectory. When reflecting on the current landscape in an exclusive interview and follow up discussion with Wellington Live, Duignan tells us that “the problem with the initial set of Chatbots that became widely available in November last year, is that they are public. The information you put into them could be used for further training.”
Elaborating on the implications, Duignan adds, “there’s a risk around security of information and intellectual property.” Some of these, he thinks “can be solved relatively easily, but others, particularly around intellectual property, will be important on-going issues.”
Generative AI’s ability to scrape relevant input and output data from diverse sources, including previous users, means it can learn and advance its skills without being explicitly programmed to do so.
What this process of self – learning means is that our information, requests, and instructions (input data) are being used in the machine learning process (training data), enabling better performance and more accurate results (output data) for the next user.
Its full circle, using feedback for improvement, which is great, but what happens when the scraping of input data for the use of output data in business contexts risks the leak of personal information or intellectual property?
The events within Samsung in March of this year exemplify these questions. Some of their staff shared confidential information with Chat GPT, like source code, confidential data, and private recordings for note conversion. This led to the company banning its use at the time. Other companies such as JPMorgan, Bank of America, and Citigroup have also banned or restricted the use of ChatGPT.
Paul Duignan believes that given the shared interest in upholding data security, AI companies will move to make ongoing security improvements to reduce these sorts of incidents. The recent data breaches experienced by Samsung serve as an illustration of a possible trajectory toward heightened security measures within AI down the line.
However, evolution comes with a trade-off: as Duignan explains, “If we look at the issue of our information being used for training data, there is no such thing as a free lunch. I think that you’ll have to pay for the solution to that problem. If people don’t pay for subscriptions to AI systems, then their information will be much more likely to be used for AI training. I suspect that in the future, IP which AI has not been able to get access to as part of its training, will become an important source of value for some organisations and businesses. I can imagine a situation where people use something like ‘You won’t find our IP on ChatGPT’ as a selling point.”
Providing controlled environments for AI tools and emphasising confidentiality will form the bedrock of AI’s safe implementation in business. ‘Hoist’, which is a NZ based software company, safeguards input, output, and training data. It allows users to tailor and train generative AI to benefit their business specifically, while protecting intellectual property and private data.
Wrapping up the discussion, Duignan drives home this crucial point in the interview by emphasizing, “This is not a reason to shy away from AI, I think businesses should be assessing ways they can use AI, but they just have to do so in a careful manner.”
In a business landscape filled with such uncertainty, the importance of secure and private data is one thing we can be sure of. While generative AI is proving its remarkable capabilities, prioritising data protection and cyber security is something all businesses need to confront during its integration.