Pitfalls of Horizontal AI and the Rise of Vertical AI Solutions

In the dynamic landscape of artificial intelligence (AI), businesses are constantly seeking innovative solutions to enhance efficiency, productivity, and overall performance.

AI solutions typically fall into two broad categories:

• Targeted, Vertical AI solutions have several advantages over broad, Horizontal AI for businesses.
• Horizontal AI, also known as General AI, attempts to provide a one-size-fits-all approach to diverse business challenges. While the concept may sound appealing in theory, the practical implementation often falls short, leading to a range of issues that hinder the effectiveness of these solutions. Let's explore some of the critical drawbacks of Horizontal AI and why Vertical AI solutions are emerging as the more tailored and efficient alternative.

Lack of Specialization

In reality, most current AI systems are narrow in scope, designed and trained for a limited set of predefined tasks within a particular industry or function, like computer vision for interacting with a web application or natural language processing for reading a claims document.

Vertical AI solutions are much more effective in the business context because they are purpose-built for the specific challenges and use cases of that industry or business process.

Data Overload and Noise

Horizontal AI solutions often struggle with handling the vast amounts of data generated by diverse industries. The sheer volume of information can lead to data overload, resulting in an excess of irrelevant or redundant insights.  This noise complicates decision-making processes and makes it challenging to extract meaningful, actionable intelligence.

Without focused domain knowledge and customization, Horizontal AI cannot achieve the same level of expertise as Vertical solutions – lacking the fine-grained understanding of industry-specific nuances, regulations, terminology and workflows needed to be truly helpful.

Vertical AI, trained on industry-specific nuances, language and processes is more capable of sifting through the noise more effectively, providing businesses with targeted and relevant information.

For example, in the Insurance space a 'Catastrophic Event' indicates a particular type of claim that requires a specific business process to handle – which a Vertical solution understands whereas a general model may see the word ‘catastrophic’ and hallucinate a story to justify the classification of ‘Catastrophic’. Vertical AI can be designed from the ground up based on a company's unique data and priorities, with a narrower and more in-depth training approach.

Training a Horizontal AI model to understand the nuances of multiple industries and tasks requires immense computational power and extensive datasets. Achieving proficiency in all areas is a monumental task, and even the most advanced systems may struggle to keep up with the evolving complexities of various sectors. Vertical AI solutions, on the other hand, can be trained more efficiently on focused datasets, ensuring a higher level of expertise and accuracy in specific domains.

Integration Difficulties

Implementing Horizontal AI across diverse business functions often leads to integration challenges.

Different departments may have disparate data structures, applications, processes and requirements, making it difficult for a universal AI system to seamlessly integrate with all aspects of an organization.  Overcoming this challenge requires significant custom engineering to optimize and align the AI solution(s) to seamlessly integrate with existing workflows, processes, regulations, systems etc.

Vertical AI solutions, designed with a deep understanding of specific industries, can offer smoother integration and greater customization improving adoption and facilitating a more streamlined and effective deployment.

Vertical AI can be both a complete system (such as our Digital Coworker) or a purpose-built modular component (such as our InsurGPT^TM component) that plugs directly into key business processes and user workflows, for faster and lower-cost deployment.

Data privacy and security are also enhanced with Vertical over Horizontal AI. Narrowly focused models only need access to relevant industry or process-specific data, reducing risks from exposing broad datasets. And specialized Vertical AI solutions are less vulnerable to distribution and misuse beyond their intended applications.

In Summary…

While Horizontal AI continues advancing as a long-term goal, Vertical AI delivers specialized intelligence focused exactly on business priorities and use cases. It allows for deeper domain expertise, more seamless integration into operations, and enhanced privacy - making Vertical AI solutions a smarter choice for most AI adoption in the near future.

As AI capabilities mature, the combination of both broad and narrow solutions will power new innovations, but for now - Vertical AI is best suited to drive real business value.

‍

Read more

Forum Ventures are creating a map of Vertical AI players across InsurTech and many other spaces (see image below).

Read more about their views on the key factors that separate Vertical and Horizontal AI solutions here.

‍

(click here to enlarge image)