"Here’s what we look for... (in an AI company): Are they doing more than basic data analysis? Are they creating their own data...? Do they use this data to create systems that constantly get smarter...? Do they have iterative technology that reduces the need for humans in the loop?" - Arif Janmohamed, Fortune.
The public's misconception of artificial intelligence is an ethereal computer based being. AI in this fictional ideology follow a trend of omniscience and of being able to save or destroy everything in existence. The more common misconception is that any data analytics or automation is AI. That sort of “Artificial Intelligence” is as intelligent as an Excel spreadsheet or an assembly line machine, which is only capable of doing a task it was designed and told to do. AI in today’s marketplace is trying to capitalize on a explosion of interest in AI and recent advancements in Deep Learning. Whether or not these companies are actually utilizing recent advancements or improving upon them is a question that remains to be seen. In our experience, most companies fail at answering any of the questions above, and are actually less capable than they claim to be.
Defining AI can be difficult because of the number of dimensions computer based systems can be seen through, the relatively new age of AI discussion and lack of computer terminology. One of the best ways of questioning and defining what AI actually is, is demonstrated in the above quote. These questions get to the complexity of what a AI really is in today's technological atmosphere. Is it doing more than data analysis? Is it creating its own data? Is is getting smarter then it was before? And more importantly does it reduce the need for humans by doing the work for them?
What people should think of as AI in this current computer age is automatic automation. Automation is the control of process by computer. Automatic means a device or process working by itself with little or no direct human control. Automatic automation is automation squared: where the machine is learning and teaching itself to automate and create more automation. It needs to be able to analysis data, learn to do things with that data, create its own data and continue to do this with newer and newer information. If you have two exactly the same AI but feed them slightly different information they should come up with different results.
Inline with a real AI is integration. Creating the support systems that make up the ‘body’ of an AI. AI systems integrations is the ability to connect individual software systems to a core AI system and create a common speech or communication protocols for them to talk with each other. An AI needs data for it to make decisions and it has to understand that data. To do that you can give it tools or make it a part of systems that can take data and turn it into information it understands.