The Risks of Relying on Algorithms

Published on
August 3rd, 2019
45 minutes

The Risks of Relying on Algorithms

The Interview ·
Featuring Cathy O'Neil

Published on: August 3rd, 2019 • Duration: 45 minutes

Cathy O'Neil, CEO of O'Neil Risk Consulting & Algorithmic Auditing, explains to Strategic Insight Group CEO Dee Smith why she is auditing companies' algorithms. O'Neil says it began when she moved from academia to trading during the financial crisis and realized the financial world couldn't control their algorithms. Now she has turned to corporations more generally, where she believes algorithms have become dangerous "black boxes." Her view: business leaders need rigorous methodology and tools or they will face catastrophic results that lead to litigation and serious reputational risks. Filmed on June 5, 2019 in New York.



  • PR
    Peter R.
    19 September 2019 @ 12:11
    This was outstanding. I think it is interesting how lopsided the thumbs up to thumbs down for this one is, yet how few thumbs in either direction there were compared to other videos. I know I skipped over this one initially, but ultimately came back to it and sure was glad I did. It sure would be great to find a way to get more people to view this video. Maybe Cathy can help tweak the RV algorithm 😉!
  • CC
    Charles C.
    23 August 2019 @ 23:11
    absolutely fascinating. thanks for this interview RV
  • JH
    Joseph H.
    17 August 2019 @ 18:46
    Wonderful interview. Schedule a follow up RV. John von Neumann... 'can we survive technology?' My question. Can we survive the maniacally modeling modern mathematician on Wall Street? Or anywhere else for that matter. We just may with more people like Cathy.
  • DP
    David P.
    9 August 2019 @ 06:41
    Probably the most high impact interview I heard this year. I would advise to read Homo Deus and watch this interview again. Makes me more optimistic for the future. Even at this early stage it is insane to leave the algos alone. One hopes that algo control will become a new useful job as we desperately need it, it seems. A refreshing change from the usual blindly optimistic, second order effect ignorant, narrative we usually hear from big tech.
  • WS
    William S.
    3 August 2019 @ 22:36
    Excellent explanation of how to assess Algorithm applications. You think u are doing something good when in fact maybe not. And the Canary...the regulation when something goes very wrong. So any business relying on Algos should third party verify its doing ethically, morally and business wise what is intended. Great video..
    • DY
      Damian Y.
      8 August 2019 @ 07:05
      Just out of curiosity, what sort of morals and ethics are you talking about. A left leaning vegan is going to have different morals and ethics than a right leaning guy who likes to hunt animals in his free time. Who is going to make up the third party of people and pass on their morally superior ideas onto everyone else? Ethics, morals and "business wise" is very subjective and can mean different things to different people. If it's a private business then the owner should have the right to choose who he want to work for him/her, just like a customer has the right to choose where they want to spend their money. Everyone has different ideas of what is morally and ethical right, you can't please everyone.
  • GL
    Gregory L.
    7 August 2019 @ 19:53
    Wonderful interview. She needs a badge she can flash, “Hello, I’m from the algorithm police. I’m here to stop you. “
  • NR
    Nelson R.
    7 August 2019 @ 02:37
    Not a coincidence that Cathy was attracted to music, math and music have a deep connection. ♾ 🎼
  • TS
    Todd S.
    7 August 2019 @ 01:15
    Very good interview - almost skipped this one but glad I watched. Well done !
  • VF
    Victor F.
    6 August 2019 @ 20:10
    Carl Jacobi's maxim "Invert, always invert"
  • SU
    Shakeel U.
    6 August 2019 @ 18:20
    Very insightful, superb 😀
  • MB
    6 August 2019 @ 17:59
    Thank you Raul and all the RV team for this interview! I suggest paying much more attention to algos impact on the current equity market. It’s the first time in the history of financial markets when so many transactions rely on robots and algorithms. Speaking of possible recession, in my opinion, it’s much easier to strike a deal with Citadel and other HFTs to program their robots to expand the bull run than to negotiate it with private and institutional investors. Robots can be programmed to ignore the fundamentals that Raul and other speakers constantly mention. Any thoughts on that?
  • DC
    Douglas C.
    6 August 2019 @ 17:27
    Fantastic interview. This is why I subscribe to RV. Brilliant and insightful. Well done.
  • PJ
    Peter J.
    6 August 2019 @ 08:05
    A very interesting interview
  • KL
    Ken L.
    6 August 2019 @ 04:34
    Cathy was very impressive! Great vertical domain expertise. If she hasn’t, she needs to reach out to the FDA and assist them in auditing all the clinical algo’s in diagnosing cancer and scoring risk for cancer patients. These are life and death AI tools trying to get to the clinical truth. AI in HC is already a $4B industry and that’s well beyond the internal capabilities of the FDA to monitor and assess.
  • PG
    Philippe G.
    5 August 2019 @ 19:18
    Great topic and excellent explanations!
  • AJ
    Andy J.
    5 August 2019 @ 17:29
    Wonderful not only for the content, but for how well Cathy explains and substantiates her points.
  • SB
    Sean B.
    5 August 2019 @ 12:17
    This is outstanding!
  • MN
    Michael N.
    5 August 2019 @ 01:15
  • WM
    William M.
    4 August 2019 @ 22:01
    This was very good. Thank you.
  • DS
    David S.
    4 August 2019 @ 00:19
    Excellent discussion. Ms. O’Neil is certainly competent. I am apprehensive about the use of any data organizing system without knowing its assumptions and testing its output often. Algorithms work best in science and math where the past and/or assumptions do mean something. If you really believe that a hiring algorithm is showing you reality, you are wrong. I am also apprehensive about using lawsuits to improve the system. There are always cases where this needs to be done, but it looks like a whole new cottage industry for attorneys. I can see tv ads like “If you applied to Apple for a job and did not get hired, we will force Apple in court to prove that someone else was a better qualified that you were.” In like manner if you did not get a raise, a corner office… DLS
  • CW
    Christopher W.
    3 August 2019 @ 13:12
    Need more of these two. Great discussion
  • DH
    Daniel H.
    3 August 2019 @ 08:01
    With computing it is always true that garbage in, garbage out. She makes a compelling case that understanding the model behind the algorithm is critical. Otherwise you may just be kidding yourself.