a

Lorem ipsum dolor sit amet, consectetur adicing elit ut ullamcorper. leo, eget euismod orci. Cum sociis natoque penati bus et magnis dis.Proin gravida nibh vel velit auctor aliquet. Leo, eget euismod orci. Cum sociis natoque penati bus et magnis dis.Proin gravida nibh vel velit auctor aliquet.

  /  Project   /  Blog: Top-10 Reasons for Artificial Intelligence Failures – Cure Worse Than Cause – Telecom Reseller

Blog: Top-10 Reasons for Artificial Intelligence Failures – Cure Worse Than Cause – Telecom Reseller

Business graph with arrows tending downwards

By Thomas B. Cross @techtionary

The idea that AI will cure all your woes is truly mad.  AI is more data, not necessarily more useful data.  AI might work if everything and I do mean everything was constant changing from demographics, customer tastes, technology, business trends, and more.  AI might work on a single slice of the pie but not the whole one.  AI might work if you are only trying to solve a very narrow singular problem, such as how to market one kind of soup to one kind of audience in one kind of market.  Otherwise, the variable would confound even a quantum computer.

Here are some of the key AI failures:

– Too narrow – one SKU not others

– Too wide – breadth not depth – can analyze ocean water but not what’s in the ocean

– Too complex – helps with auto but not truck claims

– Costs too much – analyzing data takes a lot of time and computing

– Changing needs – AI systems work best with solving a large problem that don’t change

algorithm

– Sequential failures – one author said that AI could be used with voice-driven virtual assistants such as Siri, Alexa, Cortana and others. AI works well on general requests but not sequential failures where one question leads to another and another.  An example might include “after I go down Pearl Street, what happens when it stops at the Pearl mall?”

– Escalation failures – the AI system may provide call routing to the right agent after a long series of questions but may not provide the agent with sufficient information to complete the request doubling costs or more and reducing customer expectations and experience.

– Contraindication – one of the greatest threats to healthcare is all the “side effects” facing patients compounded with all the other “side effects” from taking often two or more other medications.  The challenge is that each human is different, and medications can impact each one differently.  One recent newscast mentioned that allergies change as we age and differently for men and women further compounding AI analysis.

– Future – AI is machine programmed for now and while models can be built as seen in weather forecasting, they are often wrong and not capable without humans to change as the climate, business and other needs do, nor address calamities and other crises.

AI systems will not solve marketing solution based on emotional impacts such as “Just Do It.”  I don’t think that Nike even understands what that means but humans might though machines likely not.  Just ask Alexa, Siri or your other favorite app to explain it and if you think it’s even close, you can call me a monkey.  B2B marketing is more challenging as analysis of terms such as new, first, best, world-class, award-winning are just “puff” pieces that mean nothing to buyers.  In B2B focus on ROI, TCO, time-saving, labor-saving, cost reduction, simplification, automation, integration and other terms that technical buyers are looking for.  If you are looking for an emotional charge to your B2B messaging try “Show Me The Money” AI solutions or anything having to do with cost-savings, headcount reductions and other key impacts.

Building AI Expert Systems

In simple terms, an expert system (ES) is a computer program or system that organizes knowledge within rules or procedures to solve problems for a particular problem or task. If properly designed and maintained, an expert system can perform at or near the level of a human expert. The key issue is that an expert system is a machine. Current systems often have the constraints of the background and limitations of its creator-designer, and the skill and knowledge of the person who uses it. Presently most expert systems fail because (1) they require too much expertise from the user—it takes an expert to use an expert system—or (2) they solve only certain classes of problems—help you make chicken gumbo soup but not cream of chicken soup. An expert system must have a diverse background reference to be effective, as opposed to an incredible ability to be efficient. Expert systems reflect the rule-based side of decision making; mathematical models, formulas, algorithms, and heuristics can easily be applied, allowing expert systems to be developed and efficiently utilized. However, the key point is that a great algorithm applied against bad or biased data will only make the outcome worse. Where management or business procedures dictate a certain realm of finite possibilities to the decision makers, an expert system can be a vital management tool. Expert systems are more like productivity aids than truly intelligent software systems. They are tools that help managers improve the flow of information. These AI-assisted “power tools” provide an effective means for improving understanding, problem-solving, or decision-making. These capabilities suggest that a wide range of expert systems will be developed to guide clerks using payroll systems, help engineers with design, and aid doctors in diagnosis (including medical contraindication analysis and malpractice risks). A manager might also use such a tool to develop new models for organizational development, training, and policy analysis. These systems service the areas of business, computing, engineering, finance, geology, manufacturing, medicine, resource management, and science. The expert systems did everything from providing estate planning and investment advice to selecting auditing procedures, configuring computers, diagnosing infectious diseases, and assessing problems with oil wells. In comparing expert systems with other knowledge technologies, the term “expert system” is used to describe rule-based technologies where the information is packaged or premixed. The term knowledge network is used to describe a network of people in a thought-processing system. There are writers who use the term knowledge engineers synonymously with expert system engineers. To define a point of reference and to signify the role of people in the process and the resulting differences, knowledge networking refers to human-based activities. This is not to say that expert systems are nonhuman. The difference is subtle. In an expert system, experts compose (better term than just program) their knowledge into a computer.

Summary – I may be an expert on AI but also humble enough to know that I really am student of where AI has been, where it is now and what it needs to be in the future.  As such treat AI with respect knowing that each requires an enormous effort, research and patience to do either well, if at all.  However, I do have skills to those I help and look forward to helping you should you need it.  My professional recommendation is to integrate your AI marketing “inside” of your product/service solution.  In others, build marketing as part of your technology in order to gain insights into how your customers are using them.  From that vantage you can adjust, redesign, facilitate the customer experience.  You can then add rewards, incentives and other ways to increase their long-term ROI value.

I recommend two activities – get smart and then get an evaluation of your effort.  Click on either to get started.

ai cert

ai eval

 

 

Source: “artificial intelligence” – Google News

(Visited 2 times, 1 visits today)
Post a Comment

Newsletter