As with any project, success depends on a number of fundamental skills and efficient processes. These are areas such as strategy articulation, getting project approval, planning, project management, and customer validation. However, AI projects also require a number of new skills:
- Understanding of new AI capabilities
- Knowledge of available AI services
- Working with AI platforms
- Accessing data sources and sensors
- Machine Learning
- Neural Network setup, configuration and optimization
- AI – Human interfacing
AI has the power to radically transform a company’s entire approach to products and way they are offered. This requires a very open mindset regarding the company’s business objectives and core strengths. It is vital that the company strategy and product planning team have people with excellent insights into the potential of AI, and new capabilities enabled by it. External expertise can be of big help for guidance on the maturity and applicability of new AI to company goals.
When it comes to AI basic services, it makes little sense to develop internal expertise on AI services, like speech recognition, translation, or smart mail response services, as these basic services are available as open source, or from multiple vendors. You just need to pick the service that best matches your needs. Major AI players like Google, Amazon, IBM, and Microsoft offer platforms for development of AI projects. These platforms offer many AI services, tools and consulting services, which can significantly cut down prototyping and customer validation time. AI specialist companies can be used for specific areas such as, setting up neural networks, training the AI system, and collection of data.
The design of AI systems needs a good understanding of the data needed by the AI systems for reliable operation, as data is their raw material for AI systems. Machine learning and neural networks form the core of any AI system. Machine learning needs validated data sets for training the neural network as well as reliable, up-to-date and good quality data as input for decision making. Some of the required data originate from dedicated sensors. Data scientists are the specialists in this area, and must be a part of the AI team, since they ensure that the data is appropriate, neural networks are setup and optimized, and the network is trained.
Humans have learned how to interact and communicate with humans over thousands of years. We have learned to interface with one another. AI is learning to mimic humans in these specific areas but interfacing AI with humans remains far from trivial. Even the simplest AI devices such as, the Amazon Echo or Google Assistant need excellent interfacing with humans, making the communication feel natural and human-like. I am more likely to use the Echo when my communication with it is so natural that I can hardly differentiate the experience from that of interacting with a human.