3 Reasons Human Intelligence is Critical to AI

Malinda Gagnon
June 6, 2018

AI and machine learning are all the rage, but the key is how they’re used. You can’t just throw AI at a problem. For some of the most difficult and nuanced challenges, AI needs to be taught – by humans. We’re using AI and machine learning in our platform to discover value in massive datasets, and we certainly need human input. Human input from our team and our clients.

The three main reasons AI needs humans is because AI is immature, people and machines working together provide the greatest value, and finally, true creativity is led by humans.

1.      AI is a like teenager.

Today, AI knows what it’s been taught but left to its own devices, it doesn’t always make the best decisions. Think of a 14-year-old going out into the world and left to do as they please… you get the picture. An example of AI that hasn’t met business and user needs is seen today with Facebook and Google. They are both feeling the pain from AI and machine learning failing to fit the bill.

Facebook’s AI failed to detect fake news, hate speech, discriminatory ads and terrorist propaganda. Zuckerberg referred to AI technology more than 30 times during his testimony to congress and said in the next ten years it will be a sophisticated tool to fight against these issues on its platform. (Washington Post, 4/11/18).

Last year, Facebook also shut down a project in which two AI platforms began having a conversation with each other using English words but not in a way people could understand. It had apparently developed its own shorthand.

Google has a similar issue as Facebook with offensive content on YouTube including extremist content, compromising images of children, and hate speech. Not only was the content shared on the platform, but major brands were also programmatically (and unknowingly) buying advertisements alongside this content. YouTube and the content publisher both benefited financially from this content. More than 250 brands responded by pulling their ads from the platform. 

One of Google’s most recent AI developments is Duplex, an AI assistant that sounds like a human complete with ‘umms’ and speech patterns that can make phone calls and speak with a human without the human knowing it’s a robot. It’s an AI feat for sure. The question is, when asked, will this AI “teenager” just book the table for 2, or will a confused real teenager on the other end of the phone manage to confuse the AI and take it out of previously learned parameters?

2.      People + machines.

The most value comes from people and machines. It’s not realistic or productive to debate people versus machines. Facebook and Google are now hiring tens of thousands of team members to help clean up their content and advertising platforms.

In response to Facebook’s data privacy scandal and the hiring they’re doing, Mark Zuckerberg said, “It's not all A.I., right? There's certainly a lot that A.I. can do. We can train classifiers to identify content, but most of what we do is identify things that people should look at. So we're going to double the amount of people working on security this year. ... So it's really the technical systems we have working with the people in our operations functions that make the biggest deal." (CNBC, 3/23/18)

In response to YouTube promoting offensive content, YouTube’s CEO Susan Wojcicki said, “Human reviewers remain essential to both removing content and training machine learning systems because human judgment is critical to making contextualized decisions on content.” (The Guardian, 12/4/17)

We’re building a software company and a consulting firm wrapped into one for these very reasons. Usually, these activities are separated, and that’s actually how we started. However, we quickly realized we were missing some incredibly valuable insights in our AI and ML platform that the consulting business brings. Our consulting projects are benefiting from insights derived from our software AI and ML. Our software development is benefiting from smart people matching its capabilities to client problems and providing a feedback loop for development.

It’s important to apply HI (human intelligence) to software development and technology implementation from the earliest stages. Bill Gates said, “The first rule of any technology used in a business is that automation applied to an efficient operation will magnify the efficiency. The second is that automation applied to an inefficient operation will magnify the inefficiency.”

3.      Creativity is human.

At this point in development, AI is “narrow AI” – it’s able to do specific tasks really well and often times better than humans. Those tasks include classifying images, playing chess or Go, generating safety alerts in vehicles or even calling a restaurant to make a dinner reservation.

To be creative, humans need to lead the way. Computers can create, but to be creative in a way that provides meaning to humans, wisdom must be applied. Jack Ma said it best, "Human being[s] have the wisdom. Machine[s do] not have the wisdom."  (CNN, 9/20/17)

Human-level AI that can understand and reason as a human would is still elusive. The way humans perceive, sort through unrelated thoughts and memories to reason and make nuanced decisions is extremely difficult for a computer to replicate.   

We’re creating a data decision platform. Decision making needs to be based on business growth objectives and how teams within the organization use and work with data to do their jobs and inform decision-making. Our consulting team is the “HI” that understands the business need and understands the people in an organization that makes success happen. Over time, as our platform matures and becomes more intelligent, our HI will minimize in certain contexts.