This is a first-person account of an attendee from Cypher 2019 held from 18-20 September. Views are personal.
It’s 11:30 AM on the first day of the three-day conference, and the Telegram group dedicated to CYPHER 2019 is abuzz with enthusiastic chatter. “Room 2 and 3 are too small for this kind of conference”, a participant said in the Telegram group created for this occasion. “Too small for the size of the audience”, chimes another.
Sign up for your weekly dose of what's up in emerging technology.
Room 2 and 3 have been dedicated by the organizers to “knowledge sessions” and “tech talks”, respectively. They have a capacity of 100 each, and often have attendees crowding around the doors.
In contrast, Room 1 is by far the largest (capacity ~800) and is dedicated to “keynotes and panel discussions”. Here, people with white hair tell their audience “where the field is going”, as I overheard the first keynote speaker putting it.
This is all par for the course for a field that has become harder to define with each passing year. What is one doing when one “does data science”? Is it software engineering? Is it math? Is it plain and simple automation of what business managers and marketers have anyway been doing? Or is it just hype?
Complicating the picture even more, are the salesmen. There are two kinds of sales pitches taking place – out-of-the-box “AI” products that try to eliminate the data scientist, and education companies trying to create evermore data scientists/engineers/professionals (take your pick, it doesn’t matter) to fuel a whole generation worth of disillusionment.
I speak to one of the salesmen. “Here’s the data,” he says, holding out his left hand like it’s carrying an invisible apple. “And here’s the insight,” in his right hand. “In the middle,” and by this point he has run out of hands, “there’s a bunch of processing and that’s where our product comes in.”
So his product starts right after data and stops right before insight. Cool…
Make no mistake – this product is part of the stack at the 40-member data science team at one of India’s largest conglomerates, who are also sponsoring and speaking at the event.
The delegation from this conglomerate is led by a genial and wise-looking man who speaks (in Room 1 of course) about ABCDE of data initiatives – viz. creating Awareness, understanding Business, solving Creatively, using Data and achieving Excellence.
And if that framework is the one thing you took away from the 3 days of conferencing, more power to you, because in the end, this is as good an attempt to describe the big fat elephant that is the field of data science circa 2019 as any other.
The talks themselves are peppered with references to “business users”. Here’s what I gathered were the key features of these abstract entities:
- Business users are not data scientists
- They are also not software engineers
- They, in fact, are probably not engineers of any kind
- They might be kids
- They are paying for all of this
If there were any actual business users at the conference, they were to be found in Room 2 (knowledge sessions) and not Room 3 (tech talks).
The curious thing about Room 3 was that it was markedly younger than the other two rooms, both in terms of the audience and the speakers. It also seemed to have a greater representation of women. Perhaps tech = young = new entrants in the industry = more women.
Either way, it is safe to assume that the actual work of developing AI/ML, both the software and the math, has been placed on the shoulders of the youth, which, now that I say it out loud, isn’t all that surprising, or a bad thing.
Meanwhile, in Room 1, there’s a debate about whether AI is good or evil, touching upon the usual suspects – Cambridge Analytica, Black Mirror, Elon Musk, Facebook chatbots. It charges up the crowd and is a big hit.
The ethical questions make people edgy and make them feel part of something big and important and unknown. In general, the whole Pandora’s box of explainability/interpretability of statistical black-box models is referred to in abstract terms by several Room 1 speakers, and those are the moments when the boring age-old concept of “business context” takes a back seat as the big challenge in data science, and math itself knocks on the door.
Later, I am chatting up a senior leader of a credit card company, who heads a team of some 80 data scientists, one of the largest in India. He is quick to point out that the team was operating at the cutting edge of tech – moving past random forests and implementing neural networks. He is also quick to point out that he was not really involved in the “day to day of it”. Meaning, he didn’t code and probably haven’t trained a model in a while. He is the big-picture guy, solving for the lack of talent, or for the incompetence of tech teams which makes thing harder to implement, or for the business users who understand neither tech nor stats. Let’s be in touch.
I come back into Room 1, and listen to more about “new paradigms”, and “democratization of data”, and about running random forests built-in data lakes on a cloud infrastructure.
It feels like a group of people connected only in their combined hope that their bet works out. The bet they have taken by calling themselves data (evangelist, scientist, engineer, analyst). The bet on an AI-led future, whatever such a future might hold.
And in that bet, I feel one with them.
This article is a part of the AIM Writers Programme. If you wish to write for us, email us at firstname.lastname@example.org