The AI Explainer For The Non-Tech Founder

Artificial intelligence technologies seem to have taken over the headlines and the minds of every entrepreneur out there. Get a better understanding of what it is, so you can build your app startup stronger.

Even if you’re just getting into the tech industry, you’ve probably been bombarded with the hype surrounding artificial intelligence (AI). The term is an umbrella for a variety of technologies, ranging from smarter analytics to machine learning and natural language processing. In the straightforward words of Stanford professor Andrew Ng, AI it’s a set of computer science techniques that gives your software super powers.

Through the right mix of superpowers and the added dash of some magical marketing spin powder, AI has taken over the headlines: techies and companies are debating AI’s potential and applicability, it’s future, and its plain awesomeness. You also get the peek into the darker side of the debate: what will happen to jobs in the future and are robots more dangerous than we’d like to admit?

Why everyone is hyped about AI

The truth of the matter is AI is more present in your life and in the everyday work of companies worldwide than you’re probably aware. From the voice assistant you get on your smartphone, to the smart thermostat in your house or the recommendations you get on Netflix, AI techniques are working in the background to offer you a better experience. it’s also worth considering that there are no industries where AI isn’t optimising business processes one way or another. From small-scale applicability like sorting cucumbers on a Japanese farm to adding an extra layer of analytics and recommendations to help manage the sales and customer care experience,  AI is helping solve problems faster and better everywhere.

Venture capitalists have shown their interest in AI techniques’ applicability as well. You can see this reflected in where the investments flow. CB Insights reports that the niche of machine learning startups is slowly reaching maturity. Incubators alone housed more than 300 AI startups, three times more than in 2016. Last year, AI startups across industries raised over $15.2B in funding. That’s a 141% jump in funding from 2016. Over 1,100 new AI companies have raised their first rounds of equity funding since 2016. That’s more than half the historic number of AI startups that have ever raised an equity round. We’ll soon see the AI industry sieve itself, with only the companies providing value to their targeted clients staying successful. In a few years’ time, whatever kind of digital product you’re building, AI technologies may be part of the main features that will drive you towards success.

We’ve laid the groundwork for AI’s potential, but what is its actual applicability? Let’s get the basics of these technologies understood, so we can see how AI is shaping industries all over the world.

How computers get superpowers with AI

When you talk about AI, the first thing you should think about is machine learning. You see, when programmers started writing code for softwares, what they would do is tell it step-by-step what actions it should make. A simple example: if you wanted to teach a robot how to walk around the room, you would have to decide where the robot starts moving, how much distance it has to move forward, how it should rotate, how much distance it should move forward again, how it should rotate, and so on. If you’ve ever played Lightbot, you get the gist of it. Put the robot in a different room, and all the programming you did would be useless. You’d have to start all over again – just like moving onto the next level in Lightbot.

When you use machine learning, you don’t tell a computer what to do, you’re telling it how to give itself feedback that is somehow meaningful. So following up on our little robot example and its task of going around the room: you could program it to move forward and avoid obstacles, calculate its path and walk through it again. Mind you, you would have to go through thousands and thousands of iterations until the machine actually learns what it’s supposed to do, but you’d get there in the end.

How machine learning works is that you completely change the infrastructure of how the software works. Instead of going on a step-by-step approach, you code for it a neural network that imitates the way our brains process inputs and outputs. Overlay several layers or networking, and you get deep learning: the technology that allows computers to recognise what’s in an image. Decide you want to process language instead of images, and you get natural language processing – what helps Siri and Alexa and Ok Google to understand what you’re telling them and give you pertinent answers.

In a nutshell, AI applications today focus on very narrow tasks and decisions. But together, these narrow AI-driven tasks are reshaping businesses, markets, and industries. And because you’ve got the big players (Google, Apple, Amazon, IBM, Salesforce, Oracle) working with AI and going open source with their discoveries, even a beginner can stitch together their own AI that will parse large datasets and come up with relevant information.

Michelle Zhou, an expert in the field where AI and human-computer interaction intersects, breaks down AI in three stages. The first is recognition intelligence: algorithms running on powerful computers that can recognise patterns and glean topics from blocks of text. The second stage is cognitive intelligence, in which machines can go beyond pattern recognition and start making inferences from data. At this stage, having a data analyst teach the machine what data points have more weight and which inferences are more useful helps the AI in providing more and more relevant insights for the business at hand. We’ll reach the third stage in Zhou’s framework only when we’ll be able to create virtual human beings – and we’re a long time away from that.

Today, most AI technologies are based on recognition intelligence and various degrees of cognitive intelligence. You can use these for almost anything, as long as there’s enough data to train the software and you have a specific outcome in mind. That’s how you get plant recognition apps, neural networks to identify fish, made-to-order haircare products, computer vision for farming, stronger computer vision for image diagnostics, natural language processing for your sales overview.

Machines are not that smart (yet)

Artificial intelligence technologies are amazing, but they’re still tied down by being narrow. The truth of the matter is you need both the database and the computing power to find relevant patterns. You also need to define end goals for the machine to focus on, so the insights it gains are actually useful. Give the system too much information to work with and no boundaries and you’ve got a machine that’s roaming too broadly and not returning any useful insights. The more complex the system and the nodes, the more patterns or trends will an algorithm be able to find.

When the system isn’t fed enough datasets and doesn’t get enough feedback, you get awkward image recognition patterns, chatbots that don’t understand what you’re telling them or analytics insights that have no business value. That’s why going for an incremental approach and making use of open sourced AI technologies, whether from Google or Amazon, gives your business a greater chance of setting up a successful technologies for your business

Why does this matter to you

There are plenty of places where you can apply AI technologies, in a myriad of formats. Whether you’re  just starting out as an app founder or are a seasoned entrepreneur, there’s a lot AI can do to support your startup. That’s why our first recommendation is that you don’t focus on the neat little tool AI is. As the saying goes, if you’re holding a hammer, everything looks like a nail. Instead, start from your business processes. What kind of data are you gathering? Where are your gaps and bottlenecks? What kind of tasks are repetitive? What decisions are you making without enough information? When you start with the problems, it’s a lot easier to come up with solutions.

Here are a few directions in which you can brainstorm how AI can help:

  • Through each phase of your product’s lifecycle
  • Each of your departments or core teams in their day-to-day work
  • Reaching your high-level business objectives (and better monitoring while you get there)
  • As a core functionality of your product that may help users get the most of it.

Whichever direction you choose to focus on, AI technologies can help you build autonomous systems, helping your team free up their time and apply it more creatively. It can help you understand faster what’s happening in pictures and videos, optimise complex systems, understand people using language and make predictions about your users’ behaviour or your startup’s future.  

Building an app startup comes with a mix of grunt work and strategic planning. The longer your business will be running, the more your information system will become more complex. And in today’s overcrowded industry, figuring out ways to gain insights faster, work faster and deliver faster can give you the edge you need to succeed. And artificial intelligence – in any way you might choose to use it – may just become the jetpack you need to reach the stars.

Start typing and press Enter to search

Add CommentCollapse