THE BOB JOHNSTON PODCAST
By Bob Johnston
March 8, 2021
SEASON
1
EPISODE
32
The A.I. (R)Evolution – Google's Barak Turovsky

Will A.I. eat the world, as Elon Musk and others are predicting? My guest today is Barak Turovsky, Director of Product for Google A.I.. He’s been working on “A.I.” for so long that there was no name for it during his days doing video and audio surveillance in the Israeli Military Intelligence Corps.

The numbers are, indeed, staggering and you can see how A.I. is truly taking over the world ever day…and in every way. Google Translate in and of itself went from a ho-hum product to over a billion users in just 2.5 years. That said, we discuss why Barak believes that adoption across text, speech and other forms is ‘evolutionary’ and not revolutionary.

We also discuss deep neural networks, Barak’s upbringing in Communist Russia – his father was an engineer but also a political activist and imprisoned for his activism. We hit on how his father’s experience shaped Barak’s life and career…his early days in the Intelligence Corps, the roadmap for A.I. moving forward, why less human involvement will enhance a human…and much more.

Barak Bio

Barak Turovsky is responsible for product management and user experience for Natural Language Understanding AI and Google Translate. Barak focuses on applying cutting edge Artificial Intelligence and Machine Learning technologies to deliver magical experiences across Google Search, Assistant, Cloud, Chrome, Ads and other products. 

Prior to joining Google in 2011, Barak was Director of Product in Microsoft’s Mobile Advertising group, Head of Mobile Commerce at PayPal and Chief Technical Officer in an Israeli start up. He lived more than 10 years in 3 different countries (Russia, Israel and the US) and fluently speaks three languages. 

Barak earned a Bachelor’s of Laws degree from Tel Aviv University, Israel, and a Master’s of Business Administration from the University of California, Berkeley. 

Listen to the episode on Apple Podcasts, Spotify, Stitcher, Google Podcasts, Overcast, Castbox, or your preferred podcast platform.
SELECTED QUOTES
Humans talk in imperfect language
On why we have a long way to go for A.I. to speak back to humans
Social networks can make extreme views seem like mainstream ideas
His contrarian view about social networks
EPISODE MARKERS

4:35 How did he end up working for Google in AI

Started in Israeli military intelligence corps

Back then, no one called it “A.I.” (statistical machine learning)

Video/audio surveillance use similar techniques to modern AI

Started with Google in commerce and payments


6:30 How he took a “so-so” Google Translate app to over a Billion users in 2.5 years

Google translate was always “artificial learning” but not always artificial intelligence

2016 big breakthrough on global level using deep neural networks

9:38 Statistical machine learning

Statistical techniques are used in AI, not even in a different league

Google Translate breaks what you’re trying to translate into chunks for easier processing


14:42 Born in communist Russia
His father was a political activist who spent some time in prison

Finished high school, military, and college in Israel

Moved to the US in 2004 to attend UC Berkeley business school and obtain MBA

15:35 His father spent time in prison

In late 70s, his father spent time in a Russian prison due to being a political activist

His father was a political activist and engineer

His father worked on Russian nuclear submarines


17:15 Educational decisions

Received a law degree in Israel

Highly employable in Israel due to military experience

Worked at a tech start-up in Israel

Worked full-time, did not attend lectures

Worked at telecom companies in Israel


21:10 1980s-2000s AI

AI started as a way to find information
No technical ability

Not widely available nor was it monetized

2015 Google brain made a big breakthrough

We did not have computing power to use AI prior to this


24:05 Feelings on AI

He takes a middle ground

The story of AI is not any different than any other disruptive technology

Any disruptive technology goes through a “hype-cycle”

There are always people who jump on the bandwagon in a bad way - the technology will take our jobs, machines will kill us, etc

The opposite - people who think AI is amazing yet don’t really understand the technology and its capacity

AI is enhancing human skills

Steve Jobs - humans are faster than Cheetahs due to technology

He believes AI should be regulated

Google has a principle where they will not reinforce an unfair bias

Tech companies - Google, Facebook, Amazon, Microsoft, etc have a responsibility to go above and beyond government regulations


28:42 The evolution of AI

It’s all about technology

Through the industrial revolution to the digital evolution

We didn’t have the computing power until now


29:55 Roadmap AI

2015 Google Brain developed a technique - including hardware

Proved deep neural networks could work at huge scale

Started a wave of technology breakthroughs with big tech


31:55 Transitioning from voice to text chat in E-Commerce

Even in text you can provide much more information
Less human involvement but enhances the human - the AI can recognize if you are simply looking to change a password and provide the steps to do so
He believes we will continue on our path of using technology but in a more natural way.
The technology continues to evolve, but currently only trains us how to use specific language when interacting
Machines are smart enough to understand open-ended questions and ambiguous language
He believes technology will continue to evolve to allow for a more natural engagement with computers


35:25 Voice vs text chat

Voice should still be available but the computers will be more enhanced and able to understand a human-level of language


36:15 Natural language processing adoption

Never ending improvement

AI intelligence community is not close to human interaction

We are far from general human intelligence and lacking in predictive skills

Even where we’ve made amazing progress - there’s still growth needed

Humans talk in imperfect language

A long way to go to understand open-ended human language and a long way to go to speak back to the user in a similar language

Technological breakthrough BERT - single launch and search with ambiguity

Created one of the biggest improvements in search quality in the history of Google search


38:51 Base-use/practical-use cases

Google’s philosophy of user first

Several products that are a natural fit for AI (Google Search)

Google can immediately use technologies like BERT for search

Shelly Palmer - How should I use AI? That is a more technological question than a business-related question

It is better to focus on solving a business problem than how to integrate AI

43:05 Contrarian view  

Social networking can make extreme views seem like the mainstream ideas

Everything in life is evolutionary

Ray Kurzweil - changes from 1910 - 1980 (less manufacturing and agriculture)

People are very adaptable


45:52 What’s on his browser

Ben Thompson - Stratechery

Reads a lot about AI

Reads a lot about business - AI stems from business problems

GUEST FOOTNOTES
SPONSOR:
RECENT EPISODES: