Let’s get real about artificial intelligence and machine learning

Machine learning can make patterns evident but only if the data used is clean, normalized and complete. Natural language processing (NLP) is a critical part of obtaining data from documents and notes or chats. Getting natural language is one thing. Knowing what to do with it is another

To get truly natural responses, we’ll need to improve your machine learning. With exposure to enough conversations, the computer on the other end of the line should gradually learn what the correct response should be. But that’s where the disconnect comes in, most people believe you can train a machine in days when in reality it takes months or every years for machines to completely learn their tasks. 

The same applies to artificial intelligence; everyone is singing its praises without understanding what is TRULY involved in a successful implementation. Let’s briefly discuss some of the obstacles that are ignored: Success rates, according to IDG research, 96% of organizations are hindered by data challenges and 80% experience reduced productivity as a result of, technology gaps, leadership failures, lack of strategy, confusion about ownership of data and lack of experience.



Data issues that arise with AI are:  Siloed data aka disparate data, technology complexity or legacy systems , lack of schema or metadata, lack of access to data, lack of ability to process the big data needed for AI to work, and finally lack of business buy in and talent / experience in the field.

Artificial Intelligence needs data to learn so all the above issues are a show stopper for sure if you aren’t prepared. As for the lack of talent, we experienced data nerds can mentor until we are blue in the face but practical applications and hands on experience are still the best ways to learn.

Abdul Razack, senior VP and head of platforms at Infosys, notes that another way to develop AI expertise is to "take a statistical programmer and training them in data strategy, or teach more statistics to someone skilled in data processing." Mathematical knowledge is foundational, Terdoslavich adds, requiring a "solid grasp of probability, statistics, linear algebra, mathematical optimization--is crucial for those who wish to develop their own algorithms or modify existing ones to fit specific purposes and constraints."

 

So remember before you make promises you can’t keep, machine learning, AL, NLP, etc… all require good data, communication within the team creating or designing, system compatibility, solid logical programming and MATH… It’s not just a cool buzzword and something to add to your resume or website to be deemed relevant.


Written by a @data_nerd for more check me out on LinkedIn