Digital Catapult AI Showcase - Responsible AI & AI Ethics
Check out our key takeaways from our panel session at Digital Catapult's High Growth AI Accelerator Showcase.
At AudioStack we take customer trust and navigating the risks and opportunities of AI technologies very seriously. We use Large Language Models, advanced machine learning techniques and have a well established AI team.
So we were very glad to be invited by our friends at Digital Catapult to a panel with diverse stakeholders from consultancies advising the public sector and also more public policy focused people.
Below are the key takeaways from our Founder and CTO, Peadar Coyle on yesterday's panel: "Ethical horizons: navigating responsible growth in AI-driven companies", featuring:
Chair: Chanell Daniels, Responsible AI Manager at Digital Catapult
Siddhant Chatterjee, Policy & Governance Strategist at Holistic AI
Peadar Coyle, Founder & CTO at AudioStack
Dr Zachary J. Goldberg, Ethics Innovation Manager at Trilateral Research
1. There’s a range of standards out there
There is a UK Algorithm Transparency Standard and it’s interesting to see how public sector organizations navigate the trust and safety issues. From research this was a bit too much overhead, but it’s definitely something that’s in there.
2. Concern, anxiety and misinformation
There is a lot of concern, anxiety and misinformation about the EU AI Act, and how the US and UK will diverge. I predict this will end up a bit like GDPR and we’ll end up with some divergence of regulatory regimes. The concern for this is that it’ll increase the compliance and regulatory burden for small companies.
3. A need for pragmatism
There is a need for pragmatism. It’s not reasonable for an early stage startup to have an AI Ethics professional - in most cases. So simply making sure you have diverse teams and stakeholders, listening to your customers, and having robust testing and quality assurance. At AudioStack for example we have a range of tests and quality checks about language and bias. I feel these organic and emergent tests that are designed for problems in your domain are more than enough.
4. Don't reinvent the wheel
You don’t need to reinvent the wheel either. For example at AudioStack we’re strong believers in the work by the Content Authenticity Initiative from Adobe and are participating members in that organization. This helps to verify that something is produced by AI and what systems created the media asset. This level of transparency is a great step forward.
Overall always good to visit an accelerator that was helpful to us in our early days and talk to startups working on innovative products. Thank you to Digital Catapult for hosting us!
AudioStack is the world's leading infrastructure for fully automated, scalable AI audio production. By connecting cutting edge technologies such as text to speech, music, AI-based post production or versioning, audio brands and agencies can build complex audio production workflows in a breeze. For the first time, audio assets can be created in real-time - unlocking completely new use cases that are faster, addressable at scale and more cost-effective than ever before.