Our online behaviour in this pandemic has confused AI machine learning (ML) systems according to a new MIT report. In one example, an organisation that uses ML to detect credit card fraud had to tweak their algorithm to account for a huge spike in interest in gardening equipment & power tools.
Also large swings in consumer behaviour (toilet paper one week to gym equipment the next!) exposed the brittle nature of AI algorithms and the need for human intervention.
Nvidia is hoping its latest AI research model will help us one day do away with stock photos (or perhaps bring out our inner artist?). Its still beta (this photo was its interpretation of my text ask 'strong ocean storms') but has potential.The system is designed to translate text into realistic-looking pictures, which can be edited. It combines segmentation mapping, inpainting, and text-to-image generation.https://www.nvidia.com/en-us/research/ai-demos/
Just as every industry is transforming, insurers have
been forced to speed up their digital agendas to remain competitive amid
COVID-19. Insurers that fail to optimise their processes risk becoming
Tech scale-up and ANZIIF's Insurtech of the Year, Codafication is digitising the
insurance industry with its latest cloud-based project management and business automation software, Crunchwork.
The business and project management platform manages
the end-to-end cycle of an insurance claim, speeding up and automating up to
80% of the claim’s life cycle for customers. The platform has been tried and tested on over 200,000 claims. To learn more visit: https://crunchwork.codafication.com/
A platform that helps identify and rectify bias or compliance bias in AI models, datasets and solutions? Yes it exists - two -year old Canadian startup Fairly AI also provides a digital assistant to auto-generate compliance reports and accelerate AI model adoption (as well as foster trust, transparency and ethics)https://www.fairly.ai/