Yes, it’s true, it’s very easy (for us) to train an algorithm to not discriminate against women. It’s also possible to ensure that ethical artificial intelligence can be your competitive advantage.
If you’re looking for a fresh perspective on developing ethical tech, listen to the latest conversation Gemma Galdon Clavell had with Sebastiano Toffaletti, Secretary-General at European Digital SME Alliance.
Conversation Highlights
The issue of trust is important for our clients. They want privacy and ethics to be their competitive advantage.
Saying you’d like to develop ethical technology is easy, execution is another matter.
So, the question is, how do we move from discourse and principles to actual practices? We help our clients with practical execution.
When cars were invented…
Cars initially didn’t have a speed limit, emissions limit or seat belts, roads or lights. It’s through a long period of time that we’ve developed these rules.Today, no one would buy a car without a seat belt. We expect to have speed limits.
As of late, we’ve been developing self driving cars. No one planned ahead that a human would walk outside of the crosswalk.
For all the wonderful things that technology can bring, a whole host of concerns will arise as well.
The answer isn’t balance. Instead, ask, “What do you want to achieve?
Engineers and developers believe that because technology is superior, people will have no choice but to accept it. This isn’t true.
Technologies that have the potential to change the world will never make it into the marketplace because it’s minimized social impact.
How can ethical tech help small & medium sized enterprises(SMEs)?
EU SMEs need more clarity. They need more than just principles. How do we make them practical? What does it mean to protect data?
We’re in the business of saying yes. If you want to collect health data, we’ll find a way, but we’ll find a way to do it legally.
When training AI, often times, you don’t need personal data, you just need to anonymize it. SMEs think that they need all the data possible. But it’s not true. If you do need personal data, how do we create a governance model so that you respect fundamental human rights, privacy and GDPR?
Just like you manage your customer’s money with caution. Banks have a system to keep your money safe and secure. Data isn’t like money, it’s similar in that it’s sensitive.
SMEs think that a small team, ~3-4 people, you can change the world. However, the key is to have a team of 10-15 people with diverse backgrounds to approach a problem and that’s the key to success.
How to mitigate gender bias in an algorithm?
We’re often asking engineers solving social problems. Gender bias it seems something we’re experience more openly.
Job search – if you’re a female, job offers will be different than if you’re a male.
Banks – credit algorithms think women don’t pay loans, but that’s not true either. Why is this happening? Mostly because banks have less data on women.
Insurance – Men are charged more? But that’s a policy decision, not an algorithm.
No one’s trained the algorithm.
It’s very easy to train an algorithm to not discriminate against women, but someone needs to do it.
#artificialintelligence #iot #bigdata #cybersecurity #apps
Advice for EU SMEs who are building algorithms
Yes, it’s a challenge to run a SME, we are one as well.
We also understand that ethics is a new field and have had bad experiences with firms who offered unhelpful advice.
Invite you to see ethics as an ally and an opportunity.
It has been for us because we bring something different to the table.
Ethical tech makes us better. Approach ethics as a risk assessment exercise, which is a lot cheaper in the long run. We’re moving from principles to real practices” #AI #ML #bigdata