Your round-up of the latest, greatest data stories

Hello Vicken,

Summer is finally here! (didn’t we say this last week?) And how do we know? Well, you could look out the window…or you could look at Auto Trader’s data for convertible car sales. You choose. Anyway, I digress…Bill Gates said this week that AI will accelerate innovation and make it easier to tackle climate change. The Microsoft co-founder said that although AI had played a “fairly modest role” so far, its potential will make innovation “far easier to do”. Gates said AI’s ability to develop models and understand materials better, will drive innovation not just in climate change, but also in medicine and education.

Speaking of AI in education, researchers at the University of Reading tricked their own professors by secretly submitting exam answers generated by AI. The ruse was revealed this week, in which the exam answers went undetected and yielded better grades than real students. The researchers created fake student identities and submitted unedited answers generated by ChatGPT-4 for online assessments for undergraduate courses. Of the 33 submissions, only one was flagged by markers, while the remaining answers gained higher than average grades than the students. The report authors suggested they may have used AI to write the actual report, although a spokesperson for the University of Reading said the study was “definitely done by humans”.

We’ve been busy at ODI Towers this week, launching a series of reports called Understanding data governance in AI. This series examines the literature to highlight the critical role of robust data governance frameworks throughout AI development. Firstly, we focus on defining the AI data lifecycle stages, and secondly, the actors involved and the interactions between them, with both of these areas setting the key context for our main focus on AI data governance considerations that we cover in the final report. We also published more blogs in our series exploring the future of data and AI. We examined the current state of play in AI data transparency, looked at the issue of people’s rights in the data supply chain, and how we can ensure broad access to data for training AI models.

The Data Dive podcast series wrapped up this week with a panel of experts from Hewlett Packard Enterprise, Github and the Open Data Institute giving their practical tips for data scientists, practitioners, and research and innovation teams. You can find that episode and all episodes in the series on Spotify or Apple Podcast, so give them a listen. 

And finally, join us next week as we mark London Data Week with Data Time TV! We’ll have a variety of guests exploring the role of data in art, culture and sport. We’ll be showing the Wizard of AI again and playing The Data Game with Gavin Freeguard. The extravaganza takes place on Tuesday 2 July 10:00-12:00 BST so book your free tickets now. 

Until next time… 

David and the Comms team

Learning sale

Get your exclusive offer of 20% off some of our most popular courses

Data Ethics Professional (tutored)
Use code LDW20 at checkout for 20% OFF*

Understanding Data Ethics & AI (self-paced)
Use code LDW20 at checkout for 20% OFF*

*discount code valid 1 - 31st July 2024

From the outside world

Auto Trader data suggests surge in demand for convertibles may finally have arrived

Auto Trader

A summer surge in demand for convertibles could be just around the corner as a heatwave continues to sweep the country.

Bill Gates says AI will make it easier to combat climate change but must be “used by people with good intent”

Sky news

The philanthropist and Microsoft co-founder told Sky's The World wWith Yalda Hakim that artificial intelligence would make innovation "far easier to do". But he also warned of the dangers of the technology being used in cyber attacks and political interference.

Researchers fool university markers with AI-generated exam papers

The Guardian

University of Reading project poses questions for integrity of coursework and take-home student assignments.

From the ODI

Understanding data governance in AI

This series examines the literature to highlight the critical role of robust data governance frameworks throughout AI development.

Understanding data governance in AI: A lifecycle perspective

There are significant gaps in the literature concerning data governance across the AI lifecycle, particularly in the areas of data access, documentation, and some ethical considerations.

Understanding data governance in AI: Exploring the ecosystem

We explore the broader network of data interactions and the roles of various stakeholders, highlighting the importance of collaboration among practitioners who develop AI/ML systems.

Understanding data governance in AI: Mapping governance

Current data governance frameworks fall short in covering all phases of AI from design through to deployment, leading to high-performing models in tests that may fail in real-world applications.

AI data transparency: understanding the needs and current state of play

There is very little transparency about the data used in AI systems - a fact that is causing growing concern as these systems are increasingly deployed with real-world consequences.

Policy intervention 3 – Enforcing people’s rights in the data supply chain

We must avoid an enforcement gap in the process of training, fine-tuning and testing foundation AI models.

Policy intervention 4 – Ensuring broad access to data for training AI models

Promote access to public and private data to cultivate a diverse and competitive ecosystem of AI development.

The future of data and AI

Generative AI can help close the data divide, allowing anyone, not just those with specialist skills, to publish, find, and make sense of data.

Data Dive podcast series

Welcome to Data Dive, a podcast from the Industry Data for Society Partnership.

How can we empower data communities in the era of generative AI?

How is generative AI altering the ways communities function online?

What makes participatory data initiatives successful?

New research defines a framework of success factors for participatory data initiatives.

Want to change how you receive these emails?

You can or .