YouTube hits new records for TV viewing

This week’s 5 stories include the continuing growth of YouTube as ‘TV’, Amazon’s retail technology spreading to more stores, AI use by Google and dating apps, and how (nearly) everything can be a data set these days.

There will be no newsletter next week as I will be away. Next newsletter on Monday 23rd.

YouTube has exceeded 10% of US TV viewing for the first time

In July YouTube became the first streaming platform ever to exceed 10% of total TV usage in the US, growing at 7% versus the month before, accounting for 10.4% of TV usage, according to Nielsen data. This gives it a clear lead over Netflix, at 8.4%, with Prime Video, the third largest channel trailing far behind at only 3.4%. In aggregate, streaming also hit record highs at 41.4%, up on the 40.3% in June. Expect more gains for streaming next month - I’ll try to watch out for the numbers - with the Olympics proving even more reason for viewers to turn to streaming content on their TVs.

Amazon’s ‘Just Walk Out’ tech is coming to more stores

Amazon’s retail technology is used in its own ‘Fresh’ stores, where shoppers can just grab items and walk out (having first checked in), but what is less well known is that the tech is being integrated into other stores at venues, most notably sports arenas and university campuses. In the US more NFL stadiums now have the tech, meaning that fans can quickly grab items like drinks and merchandise and return to their seats, rather than having to queue to pay. It is also being used in universities, mostly in the US, but also in the UK (Sussex) and in Australia, allowing students to grab what they need and walk out without queuing. Some places use the ‘cameras in the ceiling’ technology, and others use RFID tags.

Google is testing an AI ‘Ask Photos’ tool

Ask Photos is a new tool for Google users - in the US initially - that allows people to search through their photos using natural language queries like ‘What did we eat at the hotel in Paris?’ or ‘What is the best picture I have taken at the National Parks?’. It will then use its understanding of the question, and its understanding of the contents of the photos to answer, based on Google’s Gemini AI models. Obviously, with the hotel example it assumes that you took a picture of the food at the time, and its usefulness will depend on how often people take photos of their lives. You can also ask it to do tasks like ‘summarise what I did on my holiday’ or even ‘collate my best recent pictures for a family album. I could definitely use it as I seem to have taken about 4,000 pictures on my phone in the last year alone…

AI comes to dating apps

AI is getting embedded into more and more of the apps and services we use, either behind the scenes or overtly as new features. One good example is the dating industry which is using AI to create a sort of virtual ‘wingman’ for users - someone to advise on good icebreakers, or critique flirting techniques. These are designed to both offer better experiences, but also reduce the pain associated with online dating - trying to think of things to say and trying to keep approaches fresh each time. AI is also being used in the background to help spot scammers and bogus profiles; every industry is trying to find the right sort of fit for AI in what they do.

& finally - using AI to analyse high school yearbook photos

Everything is now a potential dataset for AI analysis. One recent example was a piece of academic research looking at nearly 15 million US high school yearbook photos between 1930 and 2010, to see how people’s appearance and dress had changed over 80 years. It did things like classify by features including suit, tie, glasses, and hair style and was able to track traits like ‘Individualism’ when more people started to look different to their peers and ‘Persistence’, which is how much people dressed like people 20 years before, breaking this down by geography and more. Reading this reminded me of two conversations I recently had around ad tech and targeting: channels like retail media are possible now because the wealth of data on shoppers that platforms have can now be analysed and made use for things like targeting ads much more easily than ever before.