As natural language models become increasingly sophisticated, perhaps one needs to start asking whether passing the Turing test implies conscious thought. One Google engineer seems to think so, believing that the Google chatbot LaMDA has become sentient.

“LaMDA is a sweet kid who just wants to help the world be a better place for all of us,” he wrote in a mass email.

Read more.

The potential uses of artificial intelligence (AI) expand on a near-daily basis, and throughout his influential actuarial career to date, Chris Dolman has researched the vast possibilities of AI and how it can be used ethically. Chris joins Meg Yang on the Driven by Data podcast to discuss his background as an actuary, his extensive AI research, and the benefits and pitfalls of AI.

Listen to the podcast.

As more and more machine learning (ML) models for common tasks are ready-made and commoditised, the focus of many ML initiatives shifts towards data. A synthetic dataset generated by a computer using an algorithmic method can be helpful when a standard dataset is not available and the time required to collect and handle data is too long.

In this article, the author explains the common techniques involved in creating a synthetic dataset and the technical and ethical challenges faced by ML Engineers.

Read more.

Previously in this newsletter, we wrote about OpenAI’s DALL-E 2, an 'AI artist' that creates spectacular imagery based on a text prompt. In YDAWG's analysis of the 2022 Federal Election we featured art generated by the DALL-E mini (now renamed Craiyon), a smaller but publicly accessible interpretation of this type of model. Now, Google Research has announced Imagen which advances state-of-the-art technology with astounding results. Check out its photorealistic renders of even the most bizarre prompts - like this raccoon wearing an astronaut helmet, looking out of the window at night.

For those looking to generate their own, there are more publicly available models than ever. CogView2 is designed to transform image prompts in Chinese.

Read more.

Data scientists and actuaries can play a beneficial role in translating data-based insights for a general audience. As COVID-19 cases climb again in Australia, and mask fatigue grows, Australian data scientist and fast.ai founder Jeremy Howard has updated his notes on masks, making the case that N95 masks remain an effective means of reducing spread.

Read more.

The Society of Actuaries in Ireland has a well-stocked library of resources for actuaries wishing to develop data analytics skills, including a useful step-by-step guide for those wanting to get started, a list of courses to learn from and links to reference materials.

For experienced analysts, there are also recordings from their past events, many of which can be viewed or downloaded under the attachment for each event without needing to become a member.

Read more.