Use AI and Insurance Insights to Make Better Decisions

One of the hottest topics in the business world today is artificial intelligence (AI), with 95 percent of global business leaders believing that generative AI is ushering in a new era of enterprise intelligence.I The rapid progression of generative AI, which is a type of artificial intelligence that has the ability to create material such as images, music or text, is already showing its potential to disrupt business operations throughout industries and processes. While this technology is still in its infancy, understanding its current and future capabilities, potential risks and basic risk management can help business leaders make better decisions.

What is Artificial Intelligence?
Artificial intelligence refers to the simulation or approximation of certain human responses via the capacity of a computer, or robot controlled by a computer, to take actions. Artificial intelligence includes computer-enhanced learning, reasoning and perception (cannot think, understand or feel)II, to varying degrees, ranging from preprogrammed devices to complex systems.

• Preprogrammed devices, such as Amazon’s Alexa and Apple’s Siri, respond to specified commands and tend to be simple and single-task-oriented.
• Complex systems adapt behavior to achieve goals or to imitate intelligent human behavior, such as artificial general intelligence, which refers to software that’s capable of learning any task or subject. Such capabilities do not yet exist.III

Machine learning (ML), a subfield of AI, involves developing algorithms and statistical models that enable computers to learn and make decisions or predictions based on data without explicit programming.

Recent headlines have highlighted the use of generative pre-trained  transformer (GPT). GPT-4 is a neural network machine learning model trained using internet data to generate any type of text. It is a natural language processing (NLP) model like Alexa or Siri that performs a wide range of language tasks.

One of the most notable examples of GPT-4’s implementation in the NLP space is ChatGPT. Developed by OpenAI, ChatGPT requires a small amount of input text to generate large volumes of relevant and sophisticated machine-generated text. For example, ChatGPT predicts the next word in a given text string based on patterns learned from the data on which it has been trained.

In March 2023, OpenAI released an application programming interface (API) for select businesses to allow them to incorporate the AI technology into their own websites and apps via plugins.IV Using these plugins, brands have been able to harness the power of ChatGPT to create chatbots that help consumers book travel, make restaurant reservations, create curated product recommendations and consider better employment options.V

In terms of risk management, we typically consider frequency and severity of perils. With respect to artificial intelligence perils, we should add velocity of evolving risk profiles, based on the explosive growth and adoption of these technologies. One example is that Chatbot market growth is projected to reach $3.62 billion by 2030, growing at a CAGR of 23.9 percent. Gartner estimates that by 2026, conversational AI deployments within contact centers will reduce agent labor costs by $80 billion.VI

Where is AI Being Adopted?
According to Alphabet/Google CEO Sundar Pichai, AI will “impact every product across every company.”VIII

Here are some examples of AI implementation. Some are currently available and some deployments are in process or development, including benefits AI can bring to the intellectual property landscape.IX

• Manufacturing: Identifying equipment errors before malfunctions
• Banking: Integrating biometrics and computer vision to authenticate user identities and process documents
• Services: Automating conversational tasks
• Healthcare: Capturing and recording patient interactions in exams using neurolinguistic  programming (NLP)X
• Technology: Writing boilerplate code
• Content creation: Generating blog posts, social media posts, targeted email campaigns and video scripts, music and artwork. These can both enhance and replace some knowledge worker tasks, including those performed by writers, accountants, architects and insurance brokers
• Data Analysis: Analyzing large data sets and synthesizing information into easily digestible formats
• Market Research: Gathering information, such as key players, products and services, about any industry
• Product Descriptions: Generating bulk descriptions for e-commerce sites, where product catalogs are frequently updated
• Search Engine Optimization: Generating copy that includes keywords and meta descriptions search engines can find when ranking pages

“With AI perils, we need to consider velocity as well as frequency and severity, based on the explosive growth of these technologies. It took Chat GPT just five days after its founding to reach one million users.VII”

What are Some AI Risks?
Generative AI is only as good as the information on which it was trained. Perils for companies incorporating AI currently span across copyrightXI, trademark and name infringement,XII patent infringement, discrimination and defamation. As AI continues to move into operations, risks will extend to bodily injury and property damage. Technology error-and-omission risks due to AI are related to the potential flaws and limitations of any and every product or service that relies on AI software algorithms that may cause harm or damage to clients and other third parties.XIII

Risks include:

• Contract risks, with a focus on allocation of liability, terms of use, B2B versus B2C and others
• Tort risks (an act or omission that gives rise to injury or harm to another and amounts to a civil wrong for which courts impose liability). Liability attribution is not clear: developers, users, vendors, distributors, consultants, AI systems and others
• Regulatory developmentsXIV
• Intellectual property rights (traditional patents and copyrights may struggle to adapt to the innovative and collaborative nature of AI and ML development)XV
• Deceptive trade practice risksXVI
• DiscriminationXVII and defamationXVIII
• Malfunction and errors (including disinformation, unintentional mistakes and validation risks)
• Cyber-security risks
  • Privacy and data breachesXIX (AI and ML systems rely heavily on data, including personal and sensitive information, such as patient healthcare information, that is vital for the businesses using these technologies)
  • Creation of phishing email content
  • Ability to write malware code, including dynamic malware that can bypass security tools
  • Hacking and sabotageXX

Multiple lines of insurance are required to address AI perils. These include marine and trade credit, political risk (for supply chain concerns), fiduciary liability (employee benefit plans) and employment practices liability insurance (addresses biases and discrimination in employment decisions).

Existing insurance may address particular AI perils, but there are some material gaps that require a combination of revisions to existing insurance plus creative solutions. The following matrix sets forth common coverage starting points for basic available coverage (which varies for each individual risk). The general dearth of available coverage should incentivize organizations to push insurance carriers to broaden and clarify insurance coverage and to consider alternative methods of risk transfer.

How Can Organizations Make Better AI Decisions?
Eventually, organizations in nearly every industry, geography and of every size will face AI perils.XXIII Whether from internal AI implementations or from third-party vendors, partners and supply chains, there are risk management strategies to decrease the total cost of AI risks. Insurance policies address artificial intelligence perils in one of three ways:

• Affirmative coverage
• Specific exclusions
• Silent, which creates additional ambiguity

Although we are just starting the AI journey, we can leverage lessons learned from the rise of internet cyber risks and solutions, including Internet of Things’ combinations of intangible and tangible perils and solutions.XXIV