The Future of Google Image Tagging

With recent developments to the all-round functionality of AI and machine learning, many businesses across the globe have looked to implement these solutions to ease the burden on their workforce. Google is no exception and as a fervent believer in the potential of AI, they are always looking to make use of new systems to grow the functionality of their search engine. While these developments have sometimes been rocky at best, there is undoubtedly a great deal of potential for the use of AI within a search engine. One particularly precarious area is that of image tagging, where correctly identifying key factors within a picture can be massively beneficial in helping people with visual impairments as well as delivering on specific queries through image search. Through the new update, Google is trying to implement skin tone recognition as a feature within their image search, something which – when correctly enforced – can not only help to make search results more inclusive but can also help reduce bias within product marketing. Current Image Recognition While Google already has an AI image tagging system in place, it does not currently look to identify skin tone, instead focusing on labelling objects in order to discern what the subject of the image is. Having run an image of our MD at Odyssey New Media through their current vision AI, we can see that it has picked up on several things; primarily the glasses he is wearing, the laptop he is using, and maybe most important of all, that he is indeed a person (always good to double-check). We can clearly see that the foundations for AI image tagging as here, and that they are – for the most part – pretty accurate. An image of Rob run through the Google AI The MST Scale Skin tone categorisation is nothing new in the tech world, and many companies have been using the Fitzpatrick scale since its introduction in the 1970s. While it may have been a pretty innovative idea at the time, the scale was created to gauge the impact of tanning and sun burn on individuals with pale skin, something that ensured darker tones or those that did not fit within conventional standards were completely unaccounted for. Over the subsequent years, the Fitzpatrick scale was adapted to allow for six different categories of skin tone. While this helped account for more of the global population, it was still pretty limited. The MST scale (or the Monk Skin Tone scale) was introduced to improve on the idea of the Fitzpatrick scale by offering up a larger variety of skin tones across both light and dark. According to thorough research from the Harvard professor who coined the scale as well as Google’s own data on the subject, having upward of twelve categories leads to inconsistent results while less than about eight isn’t representative of a full tonal range. Taking all of this into consideration, they decided on the happy medium of ten categories. The Benefits of MST While this may seem like virtue signalling to some, there are genuine positive implications to making MST the industry standard for image recognition. Demographics play a key role in product marketing and it can sometimes be challenging to communicate directly with your target audience. Image recognition will make it easier to target specific skin tones (particularly useful for make-up brands and other cosmetic retailers) as well as making it far more likely that a business offering a diverse range of products can be quickly found. Diversity is becoming more and more important in the digital space, unfortunately, you’re still likely to see certain skin tones far more frequently than others with specific key terms (Doctor, Baby, etc…). The MST scale will make it easier to balance the search results as well as providing the option to search specifically by certain skin tones. Accuracy is probably the most useful facet of image search and finding the right pictures to match your brand or content can prove to be pretty challenging. Google has been steadily adding additional search features over the last few years in order to make their results more accurate. Implementation of the MST scale looks to be a great way to find exactly the types of images you’re looking for with a fraction of the hassle. Diversity looks to be a key area of focus for Google in the future. We might even see it become an influencing factor for SEO rankings. The world is slowly shifting to recognise the importance of diversity in day-to-day life. While representation is something that is championed within many businesses, the online world is lagging behind. A 2022 online harassment survey reported that 36.5% of people report having been bullied online at some point in their lifetime, of these, 61% were due to appearance, and 17% due to race. While Google’s new image tagging is a step in the right direction and looks to be a highly beneficial piece of kit for marketers and relevant businesses, there are some deeper societal issues that may need to be addressed in order to truly make the digital world a more accepting place. Contact Our Experts for Digital Marketing Strategy Get in touch with our team of Digital Marketing Strategy specialists who have the knowledge and know-how to fully optimise your social media and website, allowing you to enhance the number of leads and conversions your business gets.

For more information on The Future of Google Image Tagging talk to Odyssey New Media Ltd