Fireside Chat with Swami Sivasubramanian, VP, Amazon AI

We had the pleasure of hosting and moderating a fireside chat with Swami Sivasubramanian, VP of Amazon AI, yesterday in San Francisco. Swami is one of the foremost leaders in AI industry, has been with Amazon since 2005, and leads a team of “thousands”.

Overall, we covered a wide range of trends, challenges, and debates on the future of AI. It is clear that AI/ML will transform every enterprise and every industry. The content of the session was covered in part here.

In addition, I wanted to share my notes as moderator on the questions and lightning round statements, some of which we did not get to on stage. My notes from Swami’s public comments are in Italics.

Moderator Notes for Fireside Chat with Swami Sivasubramanian

What is your vision for how impactful ML is to global 1000 enterprises? Let’s say a company has 100,000 core workflows. How many are impacted, to what degree, in what timeframe?

Swami: Every company will be transformed by AI.

Andy Jassy welcomed federal regulation on facial recognition at Recode earlier this year. SF has banned it for government use in the city. Taking regulation itself aside, what is your advice for enterprises in using facial recognition?

Swami: We advise customers to use facial recognition at 99% confidence levels and should be used to make predictions not decisions.

One CTO told me yesterday that the cloud platforms are not yet “ML in a box”. There is still a lot of work for enterprises to use the platforms. What is required to get ML as easy to use as compute or storage?

Swami: ML today is early, similar to say when SQL was launched in database.

What are the next milestones in AI/ML adoption? Will there be an "iPhone moment" for consumers?

(We did not get to this question.)

Lightning Round (True or False, and 30-Second Why)

1) If it is written in python, it is called machine learning. If it is written in powerpoint, then it is AI.

Swami: False. Amazon is not a powerpoint culture.

2) AI is currently very, very stupid. (Andrew Moore, VP and Cloud AI Head, Google)

Swami: It depends.

3) Short-term AGI is a serious possibility. (Ilya Sutskever, Co-founder, OpenAI)

Swami: I am less bullish on AGI.

4) AI should not look like humans. The goal is to create smart machines, not androids. (Robin Li, CEO, Baidu)

Swami: False.

5) Major corporations should have a Chief AI Officer.

Swami: Corporations should have someone responsible for their data and data strategy, whether you call it Chief Data Officer, Chief Analytics Officer, or Chief AI Officer.

6) Another AI Winter is ahead of us.

Swami: False.