Artificial Intelligence, a Point of View
The question about AI capabilities and enablement is a challenge in not only discerning what it is, but also what is isn’t.
Artificial Intelligence (AI) means that the technology in question can
- Learn from behaviors, patterns, profiles, and copious amounts of information over time
- Develop critical thinking skills to solve problems, manage towards outcomes, create ideas, or influence via suggestion
- Learn and improve from decisions made, actions taken, and results driven over time (i.e., children)
- Infer & discern from layers of information it has learned (e.g., humour)
For “AI” to actually work, a LOT of things are required in a Big Data environment to drive the basics of heuristics helping an AI perform the above.
Here are a few prerequisites needed before investing in anything resembling “AI” for sales as an example:
1. SAMS religiously and accurately capture LOADs of information
2. A decade or more of copious, reliable, and mature data from the above is needed (the more data, the more the AI can learn from – less data, the dumber the AI. It only learns from what you give it and cannot learn from what you don’t give it. AI cannot do what humans do naturally – intuit)
3. Many, many examples of good results, bad results, cause & effect information, and logical rules
4. Pricing models, activities against those models, outcomes, reactions, and historical effects
5. Parameters to learn from and outcomes required. Even then, it is challenge without all of the above.
A Big Data Strategy and Data Architecture will get your company closer to an AI while driving meaningful value for multiple business units over time because it is set up to learn from all activities, patterns discerned, information gathered, conclusions developed, etc.
There are NO shortcuts to AI - any vendor or consultant saying otherwise is selling a superficial story lacking veracity or the ability to actually deliver.
EXAMPLE
Think of Apple’s Siri vs Google – it’s well known that Siri is “dumb” and was limited to very, very simple tasks. This is because Apple made the conscious decision to protect user privacy and not track, learn from, and adjust to every behavior (keystroke, adjustment to any setting, app used, app downloaded, activity in app, etc, etc, etc ad nauseum). Siri only learned from access to limited information – imagine teaching a child how to read without the alphabet or simple books.
Whereas Google amasses decades worth of very accurate, ridiculously pedantic information captured by millions of people (without their consent/knowledge, BTW) to help drive the AI that makes them one of the only 2 companies who actually comes close (IBM’s Watson). Google is known as the “smarter” platform because their AI has unrestricted access and exposure to all that data with plenty of time to do the things mentioned above.
NET-NET
Artificial intelligence is only as “smart” as the quality/quantity of information you give it, the parameters set for learning, access and exposure to information, the time given to learn from actions and outcomes, the boundaries you set for actions, the interventions set for mistakes, etc, etc, etc … AI is exactly like a child learning only what you teach and expose it to.
