On March 24 2017, at Tempe — Arizona a Self-driving Uber Volvo was involved in a crash with a human driven Honda CRV. Luckily, no one was injured.
Uber stopped its self-driving services immediately and launched an investigation. As per the reports, the Uber car was not at fault and it was following the rules of the road to perfection.
Nevertheless, there were witnesses who observed that if there was a human behind the wheel, this incident could have been avoided.
The technology, which Uber used in their self-driving cars, was a level 2 autonomous system that is partial driving automation. Full Driving Automation is at Level 5.
Tesla CEO, Elon Musk mentioned a couple of months back that full automated driving is still 2 years away. The regulatory compliances and the Public acceptance of Fully Automated driving system needs thousands of miles of self-driving data to prove safety.
To put things in perspective Google’s Waymo opened its self-driving vans for public rides after clocking approx. 3 million miles of self-driving on public roads. The expectation from a human driver would not be that humungous any day.
Therefore, what we expect from Artificial Intelligence is nothing short of Big & Complex. We are over-reliant on these systems and believe them to be more capable than they actually are, leading to expectation mismatch. However, there are Simpler AI systems that can still contribute a lot.
So what is AI?
One can get many explanations online. However, the one that convinced us more was “Any system which can replace an activity which till date was done only by humans”. Therefore, that assumes even ATMs as AI systems when first introduced — Debatable, but True.
Any of these inventions, which replace human intelligence (Intelligence by definition can also mean the ability to put blue balls in blue box and red ones in red box), has always been accepted as it is a mark of human’s evolution as an intelligent species. Going forward too, this cannot be suppressed.
To explain further let me bring in the perspective of online shopping.
The Intelligence involved in Online Shopping
The current online shopping infrastructure involves clicks. Imagine this infrastructure to be equivalent to a person who can just understand yes or no.
The level of detailing required in communicating through this mode is more. Meaning, the click interface will be detailed with a lot of banners, icons and drop down to cover all possible click options.
The second level of intelligence that is predominantly available is search where the search bar can comprehend most of the category related words you type. Now the options to communicate have increased and this effectively makes your life easier.
For eg. It is always simpler to type Red sleeveless Dresses than to select dresses from western wear, select red from the color option and then sleeveless from sleeve option.
The third level of intelligence is a conversation, which our team at selekt.in is currently working on. The ability to understand customer intentions/needs clearly requires back and forth communication.
Suppose if you do not like a dress, simply giving it a like or dislike is not going to help understand the reason. However, maybe if the technology can understand “Avoid checked patterns” then it actually can know your preferences better.
The Importance of Knowledge
Conversations would always require Category Knowledge. Suppose if the user were shopping clothes for party occasion, with who is she/he going would be a very pivotal information to recommend or suggest clothes better. The ability to ask relevant questions and to identify the products that suits those needs completes the conversation loop.
This capability of the system to converse and to provide advice is a big value-adding factor when it comes to online shopping.
Is Conversational AI the Best for Online Shopping?
So let us label conversational shopping assistance as Level 3 in the AI spectrum of a Super Intelligent Shopping Expert. There is still, without a doubt, a higher level of intelligence possible to achieve but that does not take away the leap of benefits that a Level 3 Shopping Assistance can provide.
In short, we are talking about a system which can understand/remember the preferences of a customer in depth, assist and guide them, thereby reducing the effort the customer has to put every time he/she comes shopping.
But conversational AI need not always be a game-changer. An article published in the MIT Technology Review last year commented that conversational user interfaces could be annoying if it takes more actions to complete an activity.
The article focused on WeChat — the Chinese equivalent of Whatsapp that implemented Chatbots as early as 2013. The case of a pizza order was considered in which a conversational user interface took 73 taps as against a simple messaging interface (involving multiple-choice buttons) which took 16 taps.
Hence, what do we infer?
What is critical is not the level of AI capability that a system has but how does it make our lives easier. Once we understand this and get our expectations right then AI becomes more consumable.
We are not sure when a fully developed human intelligence equivalent for a specific skill would be created. However, the little steps in AI will surely be influencing our way of life in a big way.
It is exciting to be in a situation when Siri never says “I don’t Know” or Google search returns only one search result — the exact result you wanted. However, that is a tougher and a more difficult peak to climb. However, the Journey by itself would be very rewarding to humanity.
Selekt.in is a chat based interactive assistant that helps in finding products suiting your needs. Based on your requirements, Selekt.in recommends a list of products with the features, specification, application etc.
For More Details Visit : selekt.in