Google’s Senior VP in charge of Search likes to say that “search is not a solved problem”.
There’s no doubt that Google moved the world a lot closer to solving that problem by indexing the world’s information and making it searchable. However, the problem it is now facing is that search is more than typing a few words in a search box. Search is full of context, and may include images, video and sound. So Google is using artificial intelligence and machine learning to continue to evolve the way we search for information online.
Last week, at their ‘Search On’ event, the search giant made a number of announcements that give some indication on where their ambitions lie.
Google Lens is critical to these plans. Broken a part of your bike, but don’t know what to call it when typing in search? Google Lens will be able to let you search visually. Then looking for instructions to fix it? Rather than taking you to static web pages, Google’s AI will be able to direct you to the best YouTube video to watch how to fix it. Google is also building a ‘things to know’ section, so they will also be serving up contextually relevant information about that bike part and how you can fix it.These are all steps in Google’s efforts to catalogue all of the world’s information and make it searchable. With the masses of data coming in from these searches, we should expect their AI and machine learning algorithms to continue getting smarter and more accurate in being able to provide the most relevant information, whatever we’re searching for.
What does this mean for Appen? It means that the data annotation industry should expect a massive
increase in demand across multiple modalities. More complicated, multimodal searches combining text and images demand “an entirely different level of contextualization that we the provider have to have, and so it helps us tremendously to have as much context as we can,” Raghavan says. “It’s a cycle that I think will keep escalating,” Raghavan says. “More technology leads to more user affordance, leads to better expressivity for the user, and will demand more of us, technically.” It also means that there is
more responsibility on the Search Giants, such as Google, to operate with
less bias. The first is technical: the machine learning models that Google wants to use to improve search have well-documented problems with racial and gender biases. They’re trained by reading large swaths of the web, and, as a result, tend to pick up nasty ways of talking. The second is that as Google begins telling you more of its own syntheses of information directly, what is the point of view from which it’s speaking? Google search, its inputs, outputs, algorithms, and language models have all become almost unimaginably complex. Without the right training and input data, Google can't expect to reach the right unbiased outcomes from its ML models.
What we know is that Appen works with 8 of the 10 largest tech companies, including Google. Could this be the beginning of the next cycle of product work that drives another expansion in Appen's top line, al la Microsoft and Amazon's smart home products (Alexa) did historically? Purely a hypothesis at the moment and competition in the data annotation space is fierce, but this is a promising sign.
Read more:
https://www.theverge.com/2021/9/29/22698504/google-search-on-event-ai-mum-google-lens-update-changesT.E.P.
(20min delay)