menu
techminis

A naukri.com initiative

google-web-stories
Home

>

Startup News

>

CTGT aims ...
source image

TechCrunch

2w

read

251

img
dot

Image Credit: TechCrunch

CTGT aims to make AI models safer

  • Cyril Gorlla, co-founder of CTGT, has established a firm that aims to enhance the safety of artificial intelligence (AI) models and aid companies in using AI more thoughtfully.
  • CTGT's auditing approach examines a model's internal concepts to identity biased outputs. It can mitigate certain errors and help establish reliability in AI models for enterprises.
  • The firm has several unnamed Fortune 10 brands as clients, including one whose facial recognition algorithm was biased towards hair and clothing. CTGT identified the bias and provided practitioners with immediate insights.
  • CTGT claims its auditing techniques are notably performant, outpacing the methods of other AI audit platforms that train 'judge' AI to monitor in-production models, according to Gorlla. He says the firm's maths-guaranteed interpretability differs from other methods as it trains fewer models.
  • CTGT is a graduate of the Character Labs accelerator and has backing from investors Mark Cuban, Jake Knapp and John Zeratsky, and Mike Knoop.
  • 'Explainable AI' could be worth $16.2bn by 2028, according to Markets and Markets. As well as CTGT, TruEra, Patronus AI, Google and Microsoft are among the start-ups and organisations developing tools and techniques to interpret model behaviour and explain AI model decision-making.
  • CTGT aims to help those larger corporations and companies requiring AI services to better deploy models that have personalised and guided API recommendations, improved framework that builds trust, 24/7 monitoring, proactive protections, and the ability to develop custom models.
  • Gorlla believes companies increasingly want safer and more trustworthy AI models, particularly in industries like healthcare and finance which are in a fast adoption phase for the technology.
  • CTGT advises on the entirety of a client's AI model lifecycle, addressing any potential safety risks from technical oversight during modelling to monitors in production.
  • CTGT’s focus in the coming months will be on building out its engineering team and refining its platform. It is just Gorlla and co-founder Trevor Tuttle at the moment.

Read Full Article

like

15 Likes

For uninterrupted reading, download the app