Information is the most valuable asset. According to legend, Nathan Rothschild earned millions of pounds sterling because he knew the Battle of Waterloo's outcome before anyone else and manipulated the London Stock Exchange accordingly.
Today, it is possible for all companies to generate information about their state, their market, and forecasts in real-time and to develop these continuously.
As described in Part 1, unlike SaaS, Artificial Intelligence is a shaping technology that will fundamentally change our economy. Therefore, AI is by definition strategic and will most likely transform your entire enterprise.
AI requires a different system architecture. For the first time, some systems use all the enterprise data to generate information. Data silos and departments that don't actively share their data are an almost business-damaging circumstance. Your organization's data architecture and the services built on it are likely to be the most valuable assets of the next decade. They will determine whether relevant product developments succeed and enable or limit access to the market.
Given the current development of algorithms, such as GPT-3, it is hard to imagine simple tasks like purchasing a new laptop or an insurance policy will still require a human's input in 5 years.
Most likely, buyers will task AI-based systems with finding the best solution for them in the vastness of the Internet and financing it in the best possible way.
Machines will buy and sell from machines and for humans. Participation in this market will depend primarily on the ability to process information. Accordingly, the path to becoming an AI-driven company is the most relevant transformation. It will keep us all busy for a long time to come.
Many preconceptions are circulating about what AI is and can do, some of which are damaging. It is worthwhile to approach this technology without fear and attempt to debunk the many existing myths.
For example, there is currently no such thing as "the one AI." AI applications today are all niche systems that excellently solve specific use cases. Generalistic synthetic intelligence in the sense of the human mind with comparable cognitive abilities is (still) a pipe dream.
It is often said that AI is expensive and requires vast amounts of data. However, it is entirely individual how much development effort an AI requires and what is feasible with the available or purchasable data. Moreover, the ROI for AI-driven systems is usually excellent. And it is possible to synthesize training data in many cases.
Conducting a pre-analysis is the only way to assess the costs and data volume needed to build a powerful AI system.
Do not rely on standard solutions from large software houses. High tech and standardization do not mix. Those who are big want to sell a lot of the same to many. Specific developments for niche products usually don't happen or don't get the attention they deserve. Besides, you pay for the brand.
As a result, you buy mediocre software for a premium price.
This problem is more so actual for all vendors that promise a platform or one-fits-all solution. For specific use cases, such as the chatbot, that may be fine. But "the one AI platform" is currently not yet feasible, as many of the technologies used are rapidly evolving.
However, it makes much sense to start from the beginning with an architecture that allows you to expand further. A curated selection of open source solutions, if you will. However, this should be continuously maintained and challenged.
Positioning the analytics department as a "service provider" for business departments and burdening it with ad-hoc requests limits your companies ability to scale intelligence.
More clearly: You hire a team to bring out your business's essence in an understandable way. This team should therefore have the opportunity to participate creatively.
This active role is crucial for your purchasing behavior because, most likely, only the analytics department will be competent to distinguish buzz words from feasible innovation.