How to apply AI at your company — Ksenia Palke // Airspace

As the CTO of your company, it is almost impossible to onlook or resist the urge to deploy AI in some form in your company. But before you make that move, is your company in the gray zone? Are you aware of ethical considerations and AI bias? Who should you hire first a data scientist or a data engineer? Listen to Ksenia Palke, Director of AI at Airspace, as she discusses applying AI in your company.
About the speaker

Ksenia Palke

Airspace

- Airspace

Ksenia Palke is Director of AI at Airspace

Show Notes

  • 03:41
    Doing good is not enough. In the pursuit of AI expect ethical pushback
    Many people want to work at a company that does good or at least doesn't do something evil but in the line of duty, you may have to collect personally identifiable data. In the pursuit of AI and new ideas, you will come to ethical crossroads.
  • 06:45
    The gray zone
    There are companies in the gray zone that unavoidably have to use personal data to deliver utility. When using AI, you have to avoid crossing over into the unethical.
  • 09:33
    AI bias
    As you adopt AI services integrated into your company, consider that the models have biases in them.
  • 12:21
    Beyond APIs, are there ethical considerations or biases that require more due diligence?
    Whether you're buying data to use or maybe using the model or fully relying on them, you have to have an honest conversation about biases that might be affecting the models.
  • 14:14
    Do you really need a data scientist?
    Many companies don't understand that they don't need to start with a data scientist. They need data engineers or someone who will build the infrastructure and later bring on a data scientist
  • 19:17
    Getting started with models and AI
    Who comes first, a data scientist or a data engineer?

Quotes

  • “Everyone in this field faces ethical dilemmas. Do I join a company where I might be making a deal with my conscience, or do I find a company that is definitely benevolent? There's a lot of gray area. What if you are using personal data for something good? What if you built something to help people by using personal data?” - Ksenia Palke

  • “There's a lot of bias in AI. If the data you train on informs the model and the data is biased, then your AI is biased. Even before building the models, it's important to think about what biases might be in my data.” - Ksenia Palke

  • “If you trust yourself, you can consider every possible bias in the data that you're using or in the way you're building the model. When trusting a third party, you just have to trust that they took care of it. And I think those are good questions to ask when you're doing a POC with a company that you might be integrating with, whether you're buying their data to use or maybe using the model or fully relying on them, to have an honest conversation about biases that might be affecting the models.” - Ksenia Palke

  • “Where I find companies delusional is that they don't understand that they don't need to start with a data scientist. They need data engineers or someone who will build the infrastructure and later bring on a data scientist or someone who will build on top of that.” - Ksenia Palke

  • “If you've never had machine learning at your company and you bring someone on, you need to give your full support to them and actually manage the change that has happened for the entire company. Because once you go through this intelligent transformation, a lot of things are going to change. And it will require educating people, building trust, and changing your workflows.” - Ksenia Palke

  • “People in data science, machine learning, or AI feel burned out, not because they work crazy hours or anything like that. It's because what they built and were excited about is not being used properly. They feel like they're not doing their job in a way that will empower them.” - Ksenia Palke

About the speaker

Ksenia Palke

Airspace

- Airspace

Ksenia Palke is Director of AI at Airspace

Up Next: