Take 100 startups and ask them “Who is an AI startup?” I am confident the majority will say they are or at least will attach AI to their narrative.
Technology plays a big role in our life whether it is Amazon recommending us what to buy, a fin-tech startup helping us to automatically invest, an HR platform matching employee with the open positions they might be most interested. At the same time [untouchable and invisible for us] the businesses are putting more efforts in AI to solve range of problems in healthcare, transportation, banking, etc. etc.
The term AI is so much overused that barely a day passes when I don’t read about an AI startup in TechCrunch or meet them during pitching events here in Armenia. Most of them are pitching themselves as “AI startups” to mold their marketing message to AI hype. In most cases these companies at the end of the day are doing data analysis and using algorithms to reach some results or are automating some processes. I am not telling that such companies do not provide value, they actually provide a huge value as data analytics itself and process automation bring enormous values to businesses, but it is not AI.
Here is the crucial difference — AI systems are becoming more intelligent through time and getting smarter by “consuming” and analyzing more data (It’s is like a kid becoming more intelligent and smart during several years as the kid is studying new things at school). Think of self-driving cars, they get more and more data every day to store the data received from sensors (pictures, voice, etc.) and improve the system’s ability to drive better. In this automotive/transportation field an authentic AI capability enables true market disruption.
Some consumer products powered by automation are putting themselves under the umbrella of AI, but all they really do is using data analytics to make repeatable and routine decisions faster. In such cases the technology does not get more intelligent over time and does not make decisions to take different actions (like turning right to avoid hitting the rock in case of self-driving car). For those companies AI has become a depreciated term that is mostly dealing with data analytics and workflow automation. These companies often throw around the word algorithm linking it with AI. But just having an algorithm that drives to certain outcomes, it does not mean it is AI.
Here some examples I met in the past few months, that are not AI startups, but they claimed ther are AI startups:
- An HR platform matching potential employees to job announcements
- A platform finding the best tour for a traveller based on the traveller’s interests
- A tool analysing bank acount transactional data for providing a credit
Startups building authentic AI products can provide very good solutions to real consumer and enterprise problems. Here are the things that should be in everyday life of the startup so that it can claim itself an AI startup.
- First of all, the startup should do more than basic data analysis.
- The startup should collect its own data from interesting sources, then create systems which use the data constantly (simply consume as we consume knowledge to become smarter) to become smarter and in its turn, the systems have a capability to collect data independently.
- Additionally those systems should be powered by technology that should reduce the human decision making portion in that cycle.
Startup founders that claim to us AI should be very deep aware of this. Also to execute an AI startup they should have deep technical experience on what it takes to build an AI system and taking a huge AI challenge.
Content Credits: Hackernoon
Image Credits: CIO