Is Artificial Intelligence a One-trick Pony?
I write blogs and I read a lot. Makes sense. My inbox is overflowing with articles about Artificial Intelligence (AI). Is everyone really using AI? Has it now become the technology that will save the human race, or is it just a one-trick pony?
The goal is to replicate human intelligence. Findings in the AI Index project estimate that modern machines’ capacity for commonsense reasoning is far less than that of a 5-year-old child. I have never met an adult, or child for that matter, whose intelligence I would like to replicate. Guess I hang out with the wrong people.
Back to the point. Researchers are still working on solving this minor intelligence problem. The typical approach is steamrolling as much data into these systems as we can. We are no closer to mimicking intelligence than we were 20 years ago, which is how long Google has been trying to figure it out.
AI was founded way back in 1956, and the use of AI, such as in a self-driving car (which tragically just killed a person) and Siri (someone help us), has the potential for huge benefits.
The prevalent business attitude is that AI will solve all problems. Well, it won’t. According to the Future of Life Institute, “Artificial intelligence today is properly known as narrow AI (or weak AI), in that it is designed to perform a narrow task (e.g. only facial recognition or only internet searches or only driving a car).” Unless your organization will be using AI for a repetitive and narrow task, it just might not be the technology you are looking for.
Now we get to the real problem. How are we going to feed our one-trick pony? Hopefully your data is squeaky clean. I would guess it isn’t. Bad data can cause unintended consequences, and inaccurate and incorrect results. AI needs vast amounts of accurate data to establish domain expertise. Some AI solutions will promise that domain expertise upfront, but beware, you still need your data and your context to achieve the objectives. And when your business changes, and it will, it is likely that you will need to retrain the system to eliminate unexpected results. In other words, it ain’t easy folks.
Poor information quality costs an organization 30 to 35 percent of operating revenues. IDC estimates suggest that only 50 percent of content is correctly indexed or metatagged, or efficiently searchable. Right from the get-go we are dooming our one-trick pony.
We solve this problem through content optimization. I know, a sales pitch, but the technology works. We generate semantic, multi-term metadata, and classify it against one or more taxonomies. Think about it. To feed the AI application, the contextual meaning of each document needs to be searched to determine its value. It can’t be done manually, as the volume of documents is too high, and human review and decision making are inconsistent, so unreliable, as well as costly.
Content optimization delivers significant benefits to the organization as a whole. Basically, the process cleanses your corpus of content and data, dealing with dark data, data of no value, duplicates, security violations, and undeclared records. It also dramatically improves search.
I say it’s a one-trick pony. What about you? AI has not yet lived up to the expectation from 62 years ago. I’ll just have to wait and see how this scenario plays out, if I live that long.
Our webinars also address the topics explored in our blogs. Access all our webinar recordings and presentation slides at any time, from our website, in the Recorded Webinars area, via the Resources tab.