Design in the Age of Artificial Intelligence
A coverage of the panel discussion organised by Design4India
Design4India is an initiative by NASSCOM to integrate Design Thinking and the Human Centric approach in startups. A panel discussion on the topic of Design in the Age of Artificial Intelligence(AI) was held on November 23rd, 2017 at Kstart in Bangalore. Staying true to its theme, participants had to apply design thinking principles to beat the traffic and reach the venue by 10 AM on a weekday! Alas, only a few were able to make it on time including the panelists. The event began late but the conversations made up for the delay. Also, the delay was a good opportunity to get to know the other participants over coffee.
Design talk by Anil Reddy
The event started with a talk by Anil Reddy, Founder & Design Director, Lollypop Design Studio. Anil shared his journey as a designer over the past 17 years. Companies are waking up to the importance of good design. Design is no longer slapping cosmetic changes before shipping the product to the customer.
Given that the theme is about design in the age of AI, Anil shared some key learnings about a product that Lollypop has worked on. It is an AI tool for sales executives. It uses a chat based interface for communicating with sales executives ranging from on-field sales personnel to the director of sales.
During the discovery phase many interesting trends started to emerge:
- 1.4 billion people use messaging apps.
- People prefer chatbots for quick emergency answers.
- 73% of people would stop using chatbots if they had a bad experience such as repetitive questioning.
- Bots pretending to be humans are creepy.
Overall that acceptance chatbots as a user interface is on the rise.
Sales executives are quite busy and need information asap. So the interface had to be designed to provide straight-forward, crisp answers. Therefore, the chatbot was given a neutral tone. While it might seem tempting to use a different tone for each persona, it adds a plethora of challenges — both design and technical feasibility. So for the first version of the app, the bot has a neutral consistent tone across all users.
Rather than providing open-ended loops, telling the users what the bot can do provides a better way of communication. The user can ask a question such as “Show me the salespeople at Adobe” and get a reply with a list of options.
The answers are formatted as cards to enhance faster consumption of information. Also the users could go back and edit their earlier query. For instance, in the above image users can click on Adobe and change the company name or click on Palo Alto and change the location. (Pardon me for the poor image quality — photos taken in a dark room always turn out bad!)
- There are some apprehensions in using bots. One of the main reason being, while conversing with a bot only the decision is shown to the user but not how it was arrived upon. To alleviate such fears, displaying the logic along with the answer will be a good practise. eg: citing the data source
- While working on technologies like AI and chatbots, it is necessary to spend a lot of time in Discovery phase. Thorough understanding of the technical platform is mandatory to design an interface that is not just easy to use but also feasible to implement.
Anil ended the session saying,
Users always interact with an interface, not codes.
Design panel discussion
Sumit Jain — Partner, KStart Capital
Ravi Shankar — Co-Founder & CEO, Active Intelligence
Atul Batra — CTO, Manthan
Apurv Anand — Co-Founder & CTO, SigTuple
Some of the key points that came up during the discussion.
What are the current trends related to AI?
- In India, around 300 million people use Whatsapp and another 200 million using Facebook messenger. Institutions like banks see this as a huge opportunity. Having a presence where the customers are already present, makes it easier to get the brand noticed. This has spurred a lot of interest and development in AI.
- A lot of work in AI is happening in enterprise and B2B market. Some of the best examples are smart replies in Gmail, Linked, etc and autonomous vehicles.
When big companies get involved, a lot of data is generated and this can be used by other companies to build on it; thereby not having to re-invent the wheel.
- There is a shift in power across organisations. The role of a Chief Innovation Officer holds more sway than a Chief Information Officer leading to more investment in R&D.
- It is only a matter of time before banks soon have a Chief AI Officer.
What are you able to accomplish today at production grade that was not possible 5 years ago?
Technologies like SaaS, open source library frameworks, etc lower the barrier of entry for new companies even in niche fields like Machine Learning (ML), AI, etc.
What are some of the challenges in AI?
Getting the ‘right’ data for training the model is difficult. Especially during Quality Assurance(QA) phase before the product is released to the market.
What are the boundaries between technologies like AI, ML, NLP, etc?
It is quite narrow and most products are a combination of many technologies.
What is the role of design/UX in AI?
Fundamentals remain the same — understand what the user wants to accomplish and make it as easy as possible.
What is the minimum threshold of data needed to design an ML /AI model?
Depends on the use case. It is always an iteration from insufficient to sufficient data.
AI is bound to take away a lot of jobs. Is India ready for it?
Cheap labor is always an edge in a country like India. While some jobs involving complex thinking might be replaced by AI, it will take more time for AI to replace the majority of the other jobs.
The session ended with a networking lunch over pizza; courtesy of the event sponsor, Facebook.
If you liked reading this article, show your love by “clapping”.
Other articles by me