The following article is a compilation of comments from a TechTalk presentation by representatives of Stickboy Creative and Vectra Digital, Felix Lluberes, Reema Bhatia, Harrison Ambs, Josh Zachritz and Bryant Jackson In September 2019.

Artificial intelligence (AI) is a machine learning algorithm that describes a system programmed to decide what to do with information on its own. It learns and gets smarter.  Anything the human mind can do, an AI system mimics it.

AI technology has been evolving for a few decades.  An example of the first wave of artificial intelligence that most people are familiar with is the recommended products feature when shopping on Amazon. The AI system made recommendations based on your buying and browsing history compared to people like you. The goal of this protocol was for shoppers to spend more money.

 

Other examples are YouTube and Netflix and the personalized recommendations for new videos and the autoplay feature. The system does not wait for your approval - it just starts playing.

 

The second wave of AI evolution applied to more industries. AI technology integrated with sales platforms were able to more accurately pinpoint and recommend when to contact a client because they are most likely to buy.  

The third wave of AI is characterized as unsupervised action. In other words the system “just does it.” The system makes its own choices.

Artificial intelligence technology is able to evolve at an accelerated pace due to the convergence of three factors: 1)cheap cloud computing, 2) outcome model (finished result) and 3) cleanliness and consistency of the data being fed in.

Large companies like Groupon and Zulily have experienced skyrocketed sales thanks in part to deployment of technology that uses artificial intelligence.  For example, an AI system can learn how to sell the maximum amount of product in the least time at the highest price. The AI system will consider all those data points via machine learning and continue improving the decisions based on what it learns.

The system can consider multiple variables (geography, time of year, weather patterns, etc.)  that is practically impossible for humans to process in a reasonable amount of time. For instance, the answer to the question above may be different in 

Ohio in January than in Florida.

 

Simple tasks such as testing different button colors on a website or different background images are smart uses for AI.  The system efficiently tests every iteration. As it tests it slowly learns what gives you the best conversion. Importantly, an AI system does not require you to script the testing protocol.

 

Mundane tasks such as monitoring daily logs and alerts can now rely on AI to identify trends which frees up human capacity to focus on higher value activities.

 

One practical use of AI in the business environment is budget management of pay-per-click (ppc) digital advertising campaigns. Unlike a typical human 9 - 5 work schedule, an AI system is able to take action on weekends, holidays, 24x7.  

 

The AI system is also capable of composing and testing creative on a scale not practical for humans. This minimizes campaign ramp up time.accelerates attaining optimal campaign performance. Offloading these tasks to the AI system allows the human to focus on account management rather than wasting time getting hung up agonizing over ad copy, etc. 

 

A national chain of pizza restaurants, Jet’s Pizza, leverages AI to manage more than 16,800 Google ads concurrently for over 400 locations.  Imagine the labor cost that would be required to have a team effectively handle this scale of campaign.

 

As artificial intelligence evolves, it is expected that the fourth wave of this technology will involve the AI system defining and testing its own goals and protocols and reporting back to the business manager.

 

Businesses can employ some best practices to be a better shopper of AI services from technology vendors. First, ask for documentation of what information is fed into the AI system.  Second, ask about how they grade results. Inspect a change log - it should be obvious a human didn't do it.

 

Challenges with AI are not all about the technology.  For example, no matter how good the technology gets, the person or organization wanting to integrate AI must really understand the objectives of the challenge they are addressing and clearly define what they are trying to do. In addition, users must identify the data sources available to feed into an AI system. Then, this data must be captured and catalogued in a normalized fashion to help with consistency.  

 

Another challenge for AI adoption is in breaking the mold of how things have been done historically.  Sometimes, without good data, the aesthetics of design trump functionality. For example, NASA is using AI for antennae design. The designs may look very different from what humans have traditionally drawn up. But with data, an AI system has Improved efficiency and enabled the structures to be made faster.

 

Public opinion presents another hurdle for AI. For instance, the technology for driverless cars and trucks that use AI and deep learning is already developed. The biggest challenge for adoption of this technology is changing people's mindsets and overcoming their resistance. Public relations campaigns that education consumers may be needed to overcome this.

 

Sometimes AI can go bad too.  In one experiment, an objective was established to build robots that run faster than living creatures.  The AI system accomplished this goal by building extremely tall robots that fell and reached their destination faster than living creatures. Mathematically it worked but was not a practical solution.

 

To address these challenges, AI may be one voice in the boardroom.  It can help leadership teams take action on insights weaned from data and patterns found within data. But there are human factors in the ecosystem that must be factored in such as ethics, social and political impacts.  

 

What does the future of artificial intelligence look like? It will require humans to coexist with a machine and be open minded to taking instructions. Humans will need to have understanding and tolerance that a machine knows better than them.  

 

With so many mundane tasks offloaded to machines, AI will free people up to change the world.