Managing the Risks of a Generative AI Deployment
By Chris Ellis, Nintex.
When any significant technological advancement first appears on the scene, it’s often greeted by apprehension and doubt.
Camera phones were seen by many people as a potential invasion of privacy while digital banking was knocked because of perceived potential security risks.
Fast forward a decade and cloud computing caused a similar reaction. Many people did not believe that data stored in the cloud could ever be as secure as data stored on-premise.
The most recent technology to cause such doubts has been generative artificial intelligence (AI). Sparked by the launch of ChatGPT in late 2022, there has been much discussion around its impact on everything from copyright and job security to the need for proactive government regulations.
People have also expressed concern about the technology’s impact on security and privacy. Generative AI enables the creation of fake news and can facilitate very effective social engineering cyberattacks.
The best approach to deployment
While it is clear that significant concerns about the technology exist in many quarters, it’s undeniable that generative AI has much to offer both organisations and individuals. To ensure deployments create meaningful benefits, there are a number of steps IT teams must take. These include:
- Incorporate effective security measures:
Phishing attacks created by tools such as ChatGPT can be much more effective as they can iterate on responses and be far more targeted at individuals. To combat this, email security providers are fighting AI with AI by identifying and blocking scam messages generated by tools such as ChatGPT. Undertaking regular internal cybersecurity validations will be important to maintain a strong privacy and security position. - Verify the accuracy of content:
Biases in large language models are, unfortunately, inevitable. However, by tuning instructions, and regularly auditing and monitoring model performance, concerns around reliability and control can be addressed. - Work to reskill staff:
Throughout history, technological innovations have displaced workers initially. However, these innovations have led to employment growth in the long term. While some job roles may be impacted, newer job roles will emerge. Employers should work to reskill staff in preparation for these opportunities. - Don’t try to start from scratch:
Making use of pre-trained AI models instead of building models from scratch can save both time and computational resources. By using training models like GPT and BERT, which provide a range of pre-trained options, organisations can reduce the overall cost of development and deployment.
The future outlook for generative AI
It’s evident that generative AI is a rapidly evolving technology, and the pace of its development is showing no sign of slowing. While there is certainly a lot of potential, it is important to be mindful of the potential risks and responsibilities associated with it.
As development continues, it’s likely AI models will become increasingly powerful and sophisticated. There will also be advances in the safeguards that are in place, such as GPT-4’s framework to reject illegal requests that have been categorised in their policy guidelines.
In the future, organisations must monitor the responsible deployment of AI while continuously innovating for growth. With a pragmatic approach, generative AI can usher in a new era of efficiency and creativity for businesses.
At the same time, however, reckless implementation risks unintended harm. The most appropriate way forward is to adopt a strategy that incorporates cautious optimism, ethical development, and proactive governance.
It’s also important that organisations audit their generative AI usage to provide transparency and ensure security of applications. Policymakers need to create forward-thinking regulations to encourage innovation while protecting society’s interests. Employees must receive support through reskilling and job elevation programs.
There is no question that the increasing abilities of generative AI will add considerable value for many organisations in the months and years ahead. By following a careful deployment strategy and putting continuous monitoring capabilities in place, organisations can make promised benefits a reality.
