Marzia Polito, the Italian who helped Google create Lens: “There are many more benefits than risks in AI”

In 1999, when she first arrived in California, he did it “because my job didn’t exist in Italy”. And even if “it’s not like that anymore”, it definitely doesn’t seem like Marzia Polito intends to go home.

His job, the one that didn’t exist in Italy twenty years ago, is to software engineer: Develop software by applying the principles of mathematics and engineering. Applying what she studied: she graduated in Mathematics in Florence, she then perfected herself in the same field at the Normale di Pisa with a thesis on algebraic geometry. When she was 27 she moved to the United States and at the California Institute of Technology she devoted herself to computer vision and machine learning.

For over 10 years with Big G

She arrived at Google in 2010 and has worked more or less continuously for the Mountain View company since then, except for a one-year break (between 2020 and 2021 she was senior research manager at Amazon Web Services). In recent years, and with particular intensity in the last couple, she and her team have dedicated themselves to the development of Lens and its latest evolutions.

Lens is that application that allows both recognize what you see in an image (a monument, a landscape, a car model and so on) and to do online searches starting from an image: “It is the basis of the function Search your Screen and of the more recent Multisearch and it is something that could not be done without artificial intelligence”, Polito explained to us when we met her at Google I/O 2023. Simplifying, it could not be done because without the calculation and processing capabilities of AI, it would be impossible to manage all the necessary data: “Neural networks and computer vision have existed for years, but it has only recently been thought of use these technologies to make the search clearer also from a visual point of view”.

Which is something that will become more and more popular and more and more will be done, so much so that when we asked her what she will be working on in the future, partly for confidentiality and partly because it will probably really be like this, Polito replied that “Lens is very important and will continue to grow, because the need to make search results accessible also from a visual point of view will continue to grow”. Also, there will be “major integration with generative AI”.

Artificial intelligence

Billions of galaxies, planets, stars. And petabytes of data. Astronomy relies on AI

by Matteo Marino

May 10, 2023

“We need attention, but we can’t give up on AI”

Obviously, as with everything related to artificial intelligences, it is also important to understand where the data used to train them comes from, to avoid that (for example) they are a source of discrimination or errors: “Our datasets are public and transparent and are also checked by human personnel to verify that our guidelines on ethics are respected and are not conditioned by prejudices”, Polito explained to us.

Despite all these precautions, there are still many voices (even authoritative ones) that raise fears and doubts about artificial intelligence, including that of British scientist Geoffrey Hinton: should we be scared? “AIs weren’t born today, we’ve been working on them for at least 20 years – Polito replied – There are negative sides, as there are in all new technologies, which is why we proceed calmly both in development and in making these tools available to people.” And yet, “the vast majority of things that AI can do or can help us do, are positive things that we couldn’t do before. And we definitely can’t give it up.”

@capoema