Speaker: Big Data Will Fuel Future Driven By Machine Learning, AI and Bots

By:
Chris Gaetano
Published Date:
May 14, 2019
Untitled design (84)

Val Steed, CEO of K2 Enterprises and a speaker at the Foundation for Accounting Education's first-ever Emerging Trends Conference on May 14, said that, currently, 2.5 exabytes of data are generated and stored each and every day. He explained that an exabyte is equal to 1,000 petabytes, which is equal to 1,000 terabytes, which is equal to 1,000 gigabytes, which is equal to 1,000 megabytes, which is equal to 1,000 kilobytes. It is estimated that all the words that every human has ever spoken in all of time are equal to about 5 exabytes. The message he gave to his audience? Data owns the future. 

"Here's the rule: Anything you put on the internet, someone is capturing to keep somewhere. All your email, all your texts, all your thises and thats. If you think you have privacy for anything on the internet you're kidding yourself. In fact, I have a Ring doorbell in my home. I love the Nest thermostat. Do you think [they are] collecting all that from us, what we set our temps at and what we do? You better believe it," he said. 

All of this information is agglomerated into "data lakes," singular centralized data stores controlled by an entity or group of entities. These data lakes, he said, tend to be shared among companies. For example, he said, the data he gives off when he flies Delta is likely shared with companies looking to sell things to people who travel a lot. However, while targeted marketing is a common use for such vast pools of information, their main value comes from how they can be used for machine learning. A great example is the rise of facial recognition software. Practical applications took a while to get off the ground, he said, because there weren't enough faces to train computers on. As time went on, however, and more people began uploading pictures of their faces, the data reached the critical mass required to effectively implement the concept. 

However, he said, machine learning isn't a new idea and is in more applications than people might think. Microsoft Excel, for example. He showed his audience a spreadsheet showing a name, an account number, and a sum for three people. Then he used the Flash Fill option (which has been around since 2013) to set a code for one of the accounts that was then applied to the other two automatically. 

"I go through an accounting system, get some transactions in, call them to their appropriate accounts, and the system says 'I got it, and unless I see an anomaly, I'll just code that.' That is just a simple example of machine learning," he said. 

People are already using much more sophisticated applications. He said that the city of Chicago recently implemented a program whereby a computer was able to guess, based on a set of metrics, whether or not someone was about to either murder or be murdered. When such an individual is brought to the city's attention (for example, a family member calling the police expressing concern), it sends two police officers, a social worker and, if relevant, a clergy member from an appropriate religion, and offers help through services such as social programs. Those that took the help, he said, have tended to have much better outcomes than those who refused. 

This is technically machine learning, but at a much higher level, which is generally known as artificial intelligence (AI). While he acknowledged there are some arguments about what is and is not AI, he decided that it should better be thought of as just very advanced machine learning because computers still lack any sense of self-awareness and cannot generate original thoughts. With this in mind, he thought that AI is best defined as "technology that enables computers to perform decision-based tasks usually accomplished by humans." 

Despite what he believes is an overhyped understanding of AI, he said it will still vastly transform the accounting field. He noted that when he was an auditor at Deloitte many years ago, he once needed to go through every loan file at a bank and make sure the signatures were in the right places. He did this all day long. Now, he said, AI programs generate a report with a summary saying that 15 percent of the loans don't have proper signatures, making the CPA more of a consultant. With this in mind, he did not take much stock in fears that these sorts of technologies will replace CPAs. 

"AI is not likely to replace accountants," he said. "They won't succumb to it, but use it, learning to audit around certain things. You won't be going through as much data anymore, and I don't think anyone will care about that."

Click here to see more of the latest news from the NYSSCPA.