It took Netflix two and a half years to reach 1 million users. Facebook did it in 10 months. Chat GPT did it 5 days.
Just as the Internet fundamentally disrupted business plans a decade ago, so, too, is generative artificial intelligence now changing the world – only at a far accelerated pace.
Management teams are now under enormous pressure to bring to their board ideas for the use of AI and ML solutions – technologies few understand — least their organization fall behind competitors. Boards, which are famously underpopulated with technology experts, now have the responsibility to oversee proposals.
While being an early adopter of AI may seem cutting edge, the AI sword can cut both ways and leave substantial scars on the balance sheet if the use is not properly calibrated. Take the case of Zillow.
In 2021, Zillow, the on-line, real estate marketplace decided it could use AI to insert itself directly in the process not as an agent but as a home buyer and a player in the home flipping industry. It created a new service, “Zillow Offers,” which made a direct offer to sellers on their homes based on AI generated data. In so doing Zillow eliminated the face-to-face interaction with agents. In the COVID era it seemed like perfect timing. Zillow offered top dollar for homes w/above market offers to seal the deals based on AI.
In less than 8 months “Zillow Offers” needed to be shut down. Zillow wound up.
unloading 7,000 houses at a steep discount took a$304 million inventory write down, eliminate 2,000 jobs (apx 25% of full staff), and suffered substantial stock losses.
Although the AI successfully tracked many variables like local market prices, time frames, neighborhoods, etc., it missed many subtle variables an experienced agent would see – the “feel” of the home and the personal matching experienced agents provide – all with real-world negative economic to the company.
In their recently published Cyber Risk Oversight Handbook, The National Association of Corporate Directors (NACD), in partnership with the Internet Security Alliance (ISA), has produced a list of questions boards of directors should consider when evaluating the use of AI in their business operations. This list is presented below:
- What is the goal for the company or organization to employ this system?
- What is the plan to build or deploy this AI or ML application responsibly?
- What type of system is the company using: process automation, cognitive insight, cognitive engagement, or some other type? Does our board and management understand how this system works?
- What are the economic benefits of the chosen system?
- What are the estimated costs of not implementing such a system?
- Are there any potential alternatives to the AI or ML systems in question?
- How easy will it be for an adversary to execute an attack on the system based on the technical characteristics?
- What is the organization’s strategy to validate dataset collection practices?
- How will the company prevent inaccuracies that may exist in the dataset?
- What will be the damage incurred from an attack on the system in terms of the likelihood and the ramifications of the attack?
- How frequently will the company review and update its data policies?
- What is the organization’s response plan for cyberattacks involving these systems?
- What is the company’s plan to audit the AI system?
- Should the company create a new team to audit the AI or ML system?
- Should the company build an educational program for its staff to learn about the use and risks of AI and ML in general?
A generation ago, the digital technology changed the fundamental business model. However, boards who were quick to adopt these technologies also placed their organizations at substantial risk by not understanding the vulnerabilities of the technology. The same is and will be true with respect to generative AI. Just as Zillow found that the human factor needed to be kept in the loop, so, too, will boards when they consider management proposals to use AI.