Doing Enterprise AI
What does it take to start AI in your organization? Do you need a start?
Let us start by looking at What AI can do currently do and get some understanding on it’s use cases.
To sum up, The AI could replicate- how we perceive through senses, how we think and how we manage our actions. To do this the AI needs to learn, transfer, predict and move.
Once we have a bit of understanding on its capabilities- the business cases could follow. But then the big question is where do we start? what capabilities should we seek?
Let us start by what capabilities we could need? we need Talent to set goals, plan and operate, Algorithms to achieve them and Data to learn and predict. and Infrastructure to store data and run algorithms
How should we strategize? Which capabilities need more investment? how do we outperform the industry? Let us look at each of them in details to understand.
Talent
what skills we need in talent?
We need talent to create scenarios, strategies and to realize our plans. We need talent do AI research, Data Modelling and Governance.
Algorithms
Algorithms could be tricky? though most of the research in this area is open and not much patented. But with open and ongoing research it becomes tricky for the talent to identify them. Once identified getting them on board is easy but to actually tweak the hyper parameters to get your goals ?
What are our goals and what do we want to learn and predict? some of those questions are based on the Data we have.
DATA
Today we can safely categorize the data into individual, process and sensor data. going ahead it would be an imperative to identify and keep the individual and personal data separately. Can algorithms learn about the correlations between and among the data? Do we need to create or acquire or process our data to meet our AI goals? though i have seen teams processing these data into system of records, insights and engagement for analytics but we are more concerned on processing them into learnings and predictions.
Given the Data and Algorithms-what are our infrastructure needs?
INFRASTRUCTURE
Data could be a real-time or historic in nature. We would need both historic and real time data to learn ? or we can learn from historic data and do our predictions from real-time? These scenarios and infrastructure requirements are totally dependent on our AI Goals.
To add them up. What Functions do we need in an Enterprise?
To bring in the difference and outperform the competition we need to start thinking AI. and how these functions integrate with our existing business.
The post is a quick fix to kick-start AI in Organizations and would require a detailed look at Scenario planning to find the best road ahead.
- How World would be in 20 Years- Best Scenario, Worst Scenario, A Neutral One?
- What are the drivers today which will impact and direct our view?
- How AI as one of the drivers will play the part?
- How our Industry would play the part in all scenarios?
- What needs to be done to be in there in all the scenarios?
Mridul K Singh
https://www.linkedin.com/in/mridulksingh/
Originally published at https://www.linkedin.com on November 24, 2017.