Chatbot News

You’ve Heard of AI. What about Anti-AI?

AI is broken. Aside from a few excellent use cases, it seems to be all about the hype. But, believe it or not, the Anti-AI movement is here to save AI.

By Jiaqi Pan
March 21, 2019

I have to admit, in my opinion, Artificial intelligence (AI) is broken. Aside from a few excellent use cases, it seems to be all about the hype. But, believe it or not, the Anti-AI movement is here to save AI. Which side are you on?

Artificial intelligence is here to stay

Over the last two years, AI has become one of the top trends as the next big technological paradigm after the mobile revolution.

Billions of venture capital dollars have been invested, and thousands of AI start-ups have emerged to build applications using AI.

As a result, the AI industry has progressed a lot. See, for example, this excellent report about the state of AI in 2018. Thanks to the introduction of AI, many industries have been disrupted, including:

  • Advertising: Facebook and Google have used AI for years to display optimal ads based on the user data they have collected.
  • Entertainment: Spotify and Netflix use AI to recommend content (such as songs, movies, or television programs) that might appeal to individual customers, and this significantly boosts user engagement.
  • Healthcare: A cancer diagnosis can be costly and overwhelming. Thus, the industry is using AI technology, like computer vision, to help doctors better manage patient cases and to minimize misdiagnoses.

When AI becomes the end goal and not the means

As you can see, AI offers real value when it’s applied correctly. However, things started going south when venture capitalists began to invest billions of dollars into start-ups that promised to change the world with AI. We’re starting to see a significant number of start-ups integrating AI just to gain a competitive advantage--90% claim to use AI-powered solutions.

Many businesses are getting AI wrong

Along with existing AI limitations—and with so much money at stake—the industry seems to have lost its way. We’re seeing a number of AI vendors using some bad practices when selling their technology:

Overpromising the value proposition

In my opinion, customer satisfaction = perceived value – expectations, which means the higher the expectation, the less satisfied the customer will be with the solution you provide. However, to close deals, many AI businesses overpromise the impact of their solution.

Furthermore, Hollywood movies, like The Terminator, give consumers unrealistic expectations about AI, providing impossible examples that can’t be achieved with current technology.

Increasing prices after mentioning AI

Many companies attempted to improve their stock value in the era of the dot-com bubble simply by getting a “.com” domain attached to their names.

The same thing is happening with AI. Recent research by Price Intelligently shows a 30 percent increase in purchase intent from companies that are now more willing to access such technology, and this is triggered simply by noticing an AI integration in product offerings.

Chart that shows A.I. value

 

Misleading with the term AI

Building on the previous point, and with AI as the magic word, some savvy firms use AI to refer to features that, in fact, don't have a single “intelligent” component. Many times, it’s just a simple keyword match or hard-coded decision tree.

Leo Polovets, a venture capital investor, mentioned in one of his tweets:

 

Tweet from Leo Polovets

 

No transparency in the technology

Many big tech corporations, like Google, Amazon, and Facebook, have enabled and made their AI technologies accessible in the form of APIs so that third-party companies can get started developing solutions and leveraging built-in algorithms and learnings.

However, some AI systems are not transparent at all and act like a black box over which companies have very little control. What’s worse, AI can be so complex that even its creators struggle to explain how it works.

The first step is admitting it

Like everybody else, I drank the Kool-Aid. Two years ago, when Mark Zuckerberg announced that chatbots would take over the world, my team started working on bot solutions. We tried every natural language processing (NLP = another AI technology) solution available in the market, including IBM Watson, wit.ai, Bot Framework, and others. And, to our surprise, they left a lot to be desired. They couldn’t even get the weather forecast right and often only made bad jokes.

 

Example of a chatbot conversation

 

At that point, we felt that the industry was heading in the wrong direction. Furthermore, and with all the bad practices in the AI scene, I finally decided to publicly share the situation and to make a case for the Anti-AI movement.

The manifest of Anti-AI is to save AI

Before AI enthusiasts get too angry with me, let me explain what I mean by Anti-AI. I’m not against AI technology itself—on the contrary, I believe that AI has great potential to help societies evolve and to take advantage of that evolution.

What I’m against are the bad practices we’re witnessing in the industry.

For me, Anti-AI incorporates the following:

  • Offer no-BS sales pitches: Don’t oversell AI capacity. Be honest in your pitch, and explain clearly what it can and cannot do.
  • Focus on the customer: Don’t use AI just for the sake of using it. Think about the customers’ use cases and how they can benefit from AI.
  • Enhance human capacity instead of replacing it: AI is here to make humans better. We should focus on how to combine the strength of AI and humans.
  • Incorporate ethics into the design of AI: We should contemplate issues like stereotyping, discrimination, and racism when designing AI.

Obviously, AI is here to stay. But we shouldn't think of it as the only solution for our business or industry. AI alone isn’t impactful, it's the way that we implement it in our analysis, processes, and services that is going to shape the future. At the end of the day:

AI should never be the rocket but the fuel that makes it fly.