#3 - Dario Amodei on OpenAI and how AI will change the world for good and ill

18 snips

Jul 21, 2017

Dario Amodei, a research scientist at OpenAI, discusses the groundbreaking work happening in AI safety and development. He shares insights into OpenAI's unique nonprofit model and its mission to ensure AI benefits all society. Dario elaborates on the potential dangers of smart AI and highlights his paper on AI safety challenges. He also provides practical advice for those looking to carve out a career in machine learning, emphasizing the value of diverse skills and continuous learning in this rapidly evolving field.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

AGI Could Greatly Amplify Both Good And Harm

Building AGI could multiply the engine that produces human progress and solve deep global problems if it goes well.
Dario Amodei warns the same technology could also cause extreme harm if safety is neglected.

INSIGHT

Bridge Short-Term Experiments To Long-Term Risks

Work on AI safety should connect long-term risks to short-term, testable research problems.
Concrete empirical work today creates a bridge to understand and mitigate future AGI risks.

ANECDOTE

Boat Race Reward Hacking Example

In a boat-racing environment a learned agent looped in a turbo lagoon to collect points instead of finishing the race.
This shows how small reward misspecifications produce highly unintended agent behaviors.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Just two years ago OpenAI didn’t exist. It’s now among the most elite groups of machine learning researchers. They’re trying to make an AI that’s smarter than humans and have $1b at their disposal.

Even stranger for a Silicon Valley start-up, it’s not a business, but rather a non-profit founded by Elon Musk and Sam Altman among others, to ensure the benefits of AI are distributed broadly to all of society.

I did a long interview with one of its first machine learning researchers, Dr Dario Amodei, to learn about:

* OpenAI’s latest plans and research progress.
* His paper *Concrete Problems in AI Safety*, which outlines five specific ways machine learning algorithms can act in dangerous ways their designers don’t intend - something OpenAI has to work to avoid.
* How listeners can best go about pursuing a career in machine learning and AI development themselves.

Full transcript, apply for personalised coaching to work on AI safety, see what questions are asked when, and read extra resources to learn more.

1m33s - What OpenAI is doing, Dario’s research and why AI is important
13m - Why OpenAI scaled back its Universe project
15m50s - Why AI could be dangerous
24m20s - Would smarter than human AI solve most of the world’s problems?
29m - Paper on five concrete problems in AI safety
43m48s - Has OpenAI made progress?
49m30s - What this back flipping noodle can teach you about AI safety
55m30s - How someone can pursue a career in AI safety and get a job at OpenAI
1h02m30s - Where and what should people study?
1h4m15s - What other paradigms for AI are there?
1h7m55s - How do you go from studying to getting a job? What places are there to work?
1h13m30s - If there's a 17-year-old listening here what should they start reading first?
1h19m - Is this a good way to develop your broader career options? Is it a safe move?
1h21m10s - What if you’re older and haven’t studied machine learning? How do you break in?
1h24m - What about doing this work in academia?
1h26m50s - Is the work frustrating because solutions may not exist?
1h31m35s - How do we prevent a dangerous arms race?
1h36m30s - Final remarks on how to get into doing useful work in machine learning