- Original entry: 8/15/2017
- Content added 5/8/2018
I continue to study this area but as of yet, not done anything directly in it, other than using SAAS systems that use AI (like X.ai).
But I continue to study this area for application in my work because like all tech, it is continuing to get more capable at far less costs. Google’s release of Tensorflow in 2015 as open source is a huge driver towards letting anyone start to build on an AI platform for their own use.
Simple put, I look at AI as teaching technology to do things for us so that we spend less time having to do them (like scheduling), or doing things for us better than we could do before (like analyzing medical data).
But teaching technology is fundamentally what software is. It is built on If…Then statements, or If/And/Or this…then this. In a crude sense, that is the basis of all software.
But AI, as I look at it, says:
- If like this…
- or, If similar to this…
- or If Like This And These Are Also Present…
then, maybe do this.
This is all fuzzy, which is what AI is designed to do, operate in the fuzzy realms where you do not have hard and fast conditions.
So, I look at what is fuzzy in my life, that requires time for me to do, which is repetitive, which if I could teach a machine to do, would increase my productivity, which means I have more time.
Or, what is fuzzy in my life, that requires constant watching and changes, which would be better handled by a machine to watch and make changes as the need arises, rather than me sitting there watching it and making the changes.
Or, what is fuzzy when it comes to interacting with my customers (they ask me questions), which can be taught to a machine to do?
Here is an excellent application of AI by a small consumer product company
If you can find data that you own which you can run AI against, that can lead to a competitive advantage: See here: https://www.wired.com/story/ai-and-enormous-data-could-make-tech-giants-harder-to-topple/
5/8/2018: See this article summarizing the different branches of AI, where they are most applicable, and use cases, and applications for consumer products.
5/8/18: AI is democratizing where it is being made available to non-AI experts to use. Consider this that I copied from the article:
The fundamental concept of machine learning is taking a big data set and applying it in a way that you can have a machine learning from positive and negative outcomes to figure out how to correlate the two. And if you talk to somebody who knows their business, it’s not really about the computer-science aspect as much as it is about asking, “Where are the natural feedback loops in the business? Where are there opportunities for catalogued positive and negative outcomes to be stored in a repository and fed into these systems?” And every business has feedback loops.
For people that don’t necessarily understand the technology, they should think about how their organization learns and realize that every place where the organization learns and every place where there’s a feedback loop likely presents opportunities to apply AI and machine-learning technologies. And there are now foundational platforms available that can allow them to do that.
Taken from this perspective, the value is in deeply knowing your business category so that you can apply AI and deep learning to it, not in being the actual programmer who builds the system.
I post what I see and do in consumer products. But I am just one person with my own perspective. I want your opinion and observations from your point of view. Please comment below so I and others can learn. Thank you!