Opinion

March 10, 2019

“Businesses are devaluing the concept of AI”

Artificial Intelligence was supposed to be 'the new electricity' but most companies are not really using AI to its full capacity, says Jeffrey Ng, chief scientist at Founders Factory.


Jeffrey Ng

4 min read

Photo credit: Jakub Steiner

In 2017 Stanford Professor and Coursera founder, Andrew Ng, declared Artificial Intelligence to be the new electricity. “Just as electricity transformed industry after industry 100 years ago, I think AI will now do the same”.  Fast forward two years… Are we seeing business and commerce upturned by intelligent machines?

The truth is that we are in a grey area. Yes, we are seeing the proliferation of AI in business as most claim to use it to automate their services. But businesses are not making the most of AI’s capability as a predictive tool. Most are simply using it for routine data classification.

Organisations shout about AI when they are just applying it to simple tasks that humans can do in less than 1 second.

By calling these basic applications of machine learning AI, businesses are muddying the debate about what really constitutes ‘artificial’ intelligence and are devaluing its concept. A day doesn’t go by when we don’t hear another organisation shouting about their use of AI, when really they are just applying it to simple tasks that we, humans, can do in less than 1 second. With so much noise around the automation of tasks, it is harder to find the companies and startups that are making very innovative uses of AI.

Advertisement

Smart use of intelligent technology delivers the most value

Automation is the lowest rung on the AI ladder. While everybody sees the value of automation in diverting human potential away from mundane tasks, such as identifying faces and sorting documents, this is not where businesses will get the greatest benefit. Ultimately, AI’s true economic value is its ability to make predictions.

AI’s true value is its ability to make predictions.

We humans have the ability, through conscious thought processes, to take in information, apply reasoning and, based on the rules we know, apply a prediction. But our brains only have a limited capacity - we can only take in a certain amount of data, we have no scale. What this means is that we may be missing out on some learnings, siloed in our individual heads.

AI is now so advanced that it can mimic our brain’s ability to scour information and spot patterns. This benefits us greatly as AI surpasses the human brain’s capacity to run through reams and reams of data, ultimately allowing it to see patterns only apparent when dealing with large data volumes. The machine learns these patterns and can then apply a “what next” prediction to be applied to a business.

Take software engineering for example. Developers go through a lengthy training process to learn to put complex unintuitive syntax and logical processing steps to perform value-delivering tasks. Kite, has recently raised $17M, to help engineers complete their lines of code, being able to suggest the right completion in the top 3 positions 67% of the time. This is a space to watch, for algorithms that can suggest two-line completions, three-lines and ultimately whole functions.

The cost of AI prediction is coming down

It is getting cheaper for startups to access AI. Large organisations like Facebook, Google, Lyst, and others are opening open sourcing their software stacks allowing startups to train their machines quicker and at lower cost.  

Startups need to be imaginative in adding their own “secret sauce” data to their AI platforms.

For example, Founders Factory portfolio company NAVA (formerly Kompass), the city-exploration app, was initially built using a leading-edge open source software library.

However, the big tech companies are less generous when it comes to allowing access to their large data collections, keeping hold of the “new oil” in order to retain a competitive edge. Startups will find it hard to compete in terms of sheer volumes of data. They will need to be imaginative in adding their own “secret sauce” data to their AI platforms if they are to build a distinctive business.