Parable of Electricity and lessons for AI

Original article was published on Artificial Intelligence on Medium

Image by analogicus from Pixabay

Parable of Electricity and lessons for AI

AI in the 2020s will grow the same way electricity in the 1920s

Electricity: A Short History

It was around the time of the American revolution when Benjamin Franklin conducted his famous kite experiment to determine the nature of electricity and made the first attempt to harness its power. Over the years came innovators and engineers like Faraday, Cavendish, Henry, Edison, Tesla that made electricity move from labs to households.

However, a much lesser famous Samuel Insull [1] was responsible for setting up the electric grid and Harvey Hubbell [2] who patented the first electric plug back in 1904.

Lessons from these electricity pioneers

Generation, transmission, distribution of electricity was a complex problem and remains to this date. Huge power generation, transmission, and distribution companies work day and night to make the electricity available through various fossil and renewable sources. However, to a consumer, the most important facet is that you plug in your device — a mobile phone, a fan, a cooker, a grinder, a 3d- printer — and use someone else’s invention perfected in alternate scenarios and apply it to satisfy your personal and business needs.

These plug and play capabilities enhance easy interaction and hide the complexities thereby making electricity ubiquitous to the extent that no one even wonders how and where it is created.

Making AI plug and play

While we are far away from a human level AI which can do anything and everything- which by the way is coming by the end of the decade if we were to believe Ray Kurzweil- we seem to have perfected the narrow AI for certain applications.

Voice, Image, Text classification, interpretation, learning, and prediction are the capabilities that have been spawned by a huge open source community. However, each time you talk about AI and machine learning, the narrative moves to complex mathematical algorithms, data-related challenges that firmly seat AI in the hands of technology teams in the organizations, and fuel the demand for specialists like data scientists and data engineers.

Democratizing AI

By providing an interface to plug and play these AI capabilities and stringing them together in the form of business workflows, we have been able to engage with the business stakeholders and helped them participate in building the AI applications without needing to understand the complexity behind each AI module.

For example, let us consider a vehicle insurance claim scenarioThe scenario currently requires the user to report the accident with the insurer. the vehicle is then towed away to a garage where the authorized surveyor would estimate the nature and extent of damage, create a survey report, take estimation from the garage and provide it to the insurance company.The insurance company would validate the paperwork submitted by the customer and use the surveyor report to decide on whether the vehicle needs to be repaired or written off.Depending on where you are in the world, this could take days to weeks and includes a lot of subjectivity from the assessor as well as the garage.If we were to rethink of doing this using AI and machine learning, we can think of it as a workflow, where the user documentation and accident photographs are uploaded to a cloud, a sequence of ML and automation steps happen and the user is provided with an assessment of costs and an insurance company recommendation for write-off or repair- all within minutes.We can write complex algorithms for achieving it or can plug and play various open source AI modules in a simple UI that can deliver the same outcome.

By breaking down the ML problem into smaller components and connecting them into a workflow by using a simple to understand GUI that does not require any coding, we have enabled our customers to rapidly build workflows, train their algorithms and deploy ML solutions.


AI of tomorrow will be as ubiquitous as the electricity of today. By hiding the complexities and enabling a simplified UI driven ML and automation workflows, we can make AI as simple as plug and play and democratize the AI usage.