Transparency will be the most important thing to avoid bias in AI systems
Posted: Tue Jan 07, 2025 8:40 am
Artificial intelligence systems continue to be a concern for many people. Now, the head of Google's AI division, John Giannandrea, has joined these concerns . But, as he explained in a conference that focused on the relationship between humans and AI , his main concern is biased data .
This is not the first time that the possibility of machines making unfair decisions has been highlighted . There are already groups of people protesting precisely about this, and about what it means to delegate to algorithms. In addition, we must take into account the secrecy that companies have regarding these algorithms .
John Giannandrea himself has stressed the importance of transparency . “It is important that we are transparent about the training data we are using, and look for hidden biases. Otherwise, we are building biased systems.” But the company algeria phone number that Giannandrea is part of is not exactly an example of transparency. Google has been hiding its algorithms for years and barely provides data on which parameters are used to offer one result over another in its search engine.
On the other hand, they are not the only ones. As xataka.com recalls , most leading technology companies hide their algorithms , with a conduct that could be described as anything but transparent. With this, companies would try to obtain a competitive advantage over the rest of the companies. But, in return, users cannot know to what extent the information offered is biased. "If we give these systems biased data, they themselves will be biased," according to Giannandrea himself.
This is not the first time that the possibility of machines making unfair decisions has been highlighted . There are already groups of people protesting precisely about this, and about what it means to delegate to algorithms. In addition, we must take into account the secrecy that companies have regarding these algorithms .
John Giannandrea himself has stressed the importance of transparency . “It is important that we are transparent about the training data we are using, and look for hidden biases. Otherwise, we are building biased systems.” But the company algeria phone number that Giannandrea is part of is not exactly an example of transparency. Google has been hiding its algorithms for years and barely provides data on which parameters are used to offer one result over another in its search engine.
On the other hand, they are not the only ones. As xataka.com recalls , most leading technology companies hide their algorithms , with a conduct that could be described as anything but transparent. With this, companies would try to obtain a competitive advantage over the rest of the companies. But, in return, users cannot know to what extent the information offered is biased. "If we give these systems biased data, they themselves will be biased," according to Giannandrea himself.