In this episode of GMA Lifestyle’s ‘Straight from the Expert,’ Sophia Senoron and Google Philippines Head of Government Affairs and Public Policy Yves Gonzalez give us a crash course on artificial intelligence (AI).
Category
😹
FunTranscript
00:00Artificial intelligence, or AI, has been a hot topic around the world for a while now,
00:05especially with how it has been widely integrated in our lives today.
00:09Along with the raised awareness for AI is also the wider debate and questions that surround the technology,
00:16including if it would eventually replace humans in the workplace.
00:21Hi, I'm Sofia Ceneron, and I will be your host for this episode of Straight from the Expert,
00:26where we will discuss AI and how it is changing our world today.
00:34We are here at the Google offices in the Philippines, and we got in touch with Sir Yves Gonzalez,
00:39who is the Head of Government Affairs and Public Policy.
00:46Hi, Yves. How are you?
00:48Good, good. Thank you for having me.
00:50Thank you for joining us, actually. You have a lot of knowledge to impart on us today.
00:55So, I'm very, very excited.
00:57Earlier, I mentioned how AI, or artificial intelligence, is a subject of discussion today.
01:03But before we dive into the complexity, can you first define what AI is?
01:08To put it simply, we think that AI is basically when humans try to teach computers how to think,
01:17or how to act, or how to solve problems just like humans can, and even better ones.
01:22So, basically, you know, you also hear about machine learning, right?
01:26So, when we talk about machine learning, it's basically the human, or the computer,
01:30can learn, actually, from data, from information, and it becomes better.
01:35Does AI have no use in the everyday life of an ordinary citizen?
01:39So, the good thing with AI now, it's been democratized, right?
01:42So, I guess, at first, AI was primarily for computer scientists, right?
01:48But now, with all these AI tools, even, you know, a sari-sari store vendor,
01:53we had a training with a sari-sari store owner on how to improve the design of their sari-sari store, right?
02:00Or even, let's say, a tricycle driver,
02:04who wants to learn how to create a good resume, right?
02:10So, things like that.
02:11So, AI now has the opportunity to improve the lives of everyday Filipinos.
02:17AI has now, especially, there's a lot of free tools now.
02:19So, before, it was expensive, but now, so many free tools are available.
02:23It's kind of refreshing to know that even just ordinary Filipinos,
02:27like you said, a sari-sari store owner and a tricycle driver,
02:30they are able to use these free tools these days.
02:36Is AI nearing human intelligence?
02:39Eve said that although AI has an advantage in certain skills,
02:43like processing and understanding millions of images and data,
02:47AI still lacks the ability to understand and comprehend human emotions and relationships fully.
02:54The collaboration between human and AI, that will always be the case, right?
02:59It's not about one being better than the other,
03:01but actually working together, collaborating, so that we can solve some of society's tough problems.
03:08Since AI is relatively new to most people,
03:10and some of them think it is hard to use and it requires technical knowledge,
03:14is this true or false?
03:16Definitely false.
03:17So, we always say that, you know, in Google, our goal is to democratize technology, including AI.
03:22So, with Gemini, Gemini is both our AI model and our large language model.
03:28We're able to democratize AI.
03:30So, you already know how to do a Google search.
03:31Now, you can enter Gemini, and basically, Gemini will give you an AI-generated answer, right?
03:37And then, there's a lot of AI that's actually happening behind the scenes.
03:41So, you know how to make YouTube videos, you know how to watch YouTube videos, right?
03:45Powering that behind the scenes, those recommendations, there's AI there, right?
03:49When you use Google Maps or Waze, there's AI behind the scenes that's telling you the route to take.
03:54So, actually, AI is now at the stage that anyone can use it.
03:58I don't think a lot of people realize that so much of AI is already integrated.
04:02I don't think a lot of people realize that so much of AI is already integrated in our daily lives without us noticing.
04:07So, for example, when you do a Google search, then you autocorrect, right?
04:10You typed the wrong word, but the results are still correct.
04:13That's AI behind the scenes that's powering that.
04:16Okay. So, even something as simple as that is already AI.
04:19Wow. So, it's definitely very easy to use because I use it in my everyday life.
04:32There are also discussions about AI being biased and unfair. Is that true?
04:43So, definitely, depending on how the AI is trained, right?
04:47So, one of the challenges of AI is to make sure that its training reduces, removes the biases and unfairness, right?
04:56Because, for example, if your training data has a bias, then, of course, if you don't control it, the bias will keep coming out.
05:03So, it's really incumbent upon the AI developer like Google to make sure that we are conscious of these known and unknown biases.
05:13So, it's incumbent upon us to really try to counteract those.
05:16And, of course, you have to be responsible in the outputs of that, not just in the training of the AI, but also the outputs that AI generates.
05:24For example, you put a lot of training data of doctors, right?
05:28And then, so, for example, if the training data is, let's say, full of pictures of male doctors, and then you ask the AI to generate a doctor image,
05:36if you don't correct the bias, what will come out of it is full of male doctors, right?
05:42So, you all, as a company, you have to be conscious of the possible potential biases of the training data
05:48such that the output, you will consider if it's right that it's full of male doctors, right?
05:54This is a problem that has existed, and we have to solve for it, right?
05:57So, for example, when you use Gemini to generate an image, in that image, it has a fingerprint.
06:04We call it Synth ID.
06:06That fingerprint basically tells Google, Google Search, and then even the users that this image was AI-generated, right?
06:15So, we will do our part to inform people if there is an image generated in that image or if it was used by AI, right?
06:23On YouTube, for example, we require our YouTube creators that if they have used AI in a content,
06:32and the content is realistic, and then it has a possibility of deceiving someone,
06:38we require, we put a label that says there, synthetic or manipulated, or basically AI-generated, right?
06:44So, again, being able to tell people whether AI was used, especially when there are risks involved,
06:50it's part of our responsibility, it's part of our AI principles, right?
06:53So, again, a lot of these problems are long-standing, but now AI adds a new dimension to it.
07:00Since we are at a time where AI is rampant and widely used even in our daily lives,
07:07it is also important for us, the consumers, to carefully examine whether the content we interact with is AI-generated or not.
07:16What are the limitations and disadvantages of AI?
07:22That's a good question. So, especially now, AI in a sense, especially generative AI,
07:27if you think about it, we're at the infancy, this is just the beginning, right?
07:33Although, the tools we've seen are great, we have to be conscious that, again, number one, it's a bias, right?
07:38We have to be conscious that sometimes some AI tools might have some biases,
07:44might reflect some of the biases of society, and we have to be conscious of that, right?
07:48Number two, we have to be conscious that AI can make mistakes.
07:51Not because it's AI, it doesn't make any mistakes, right?
07:55So, again, it is relatively new technology, infant technology, so it can still make mistakes.
08:01But we cannot, in a sense, we shouldn't 100% rely on the output that AI gives to us, right?
08:07So, for Google, for example, in Gemini, if it gives you an answer, we have a Google It button there.
08:12So, in a sense, what does the Google It button do?
08:14It allows you to, in a sense, Google the sources that the AI used for the answer,
08:19such that you yourself can validate if the output is correct, right?
08:24That's actually really helpful, because even when you were studying back in the day,
08:28you would always have to reference the authors or whatever source you got it from.
08:32So, AI does the same thing.
08:33Correct. So, yes.
08:35That's wonderful.
08:36So, if you're just, let's say, it's a personal curiosity, then you can basically ask Gemini a question,
08:42it will give you the answer.
08:43But if you're presenting it to your boss, to your professor, to people that you need to have 100% correct data,
08:50so while AI will help you with the initial work,
08:53at the end of the day, the ultimate responsibility for the accuracy of your report,
08:58which was assisted by AI, still falls on you, right?
09:00So, again, you will still check the facts, you will check the sources,
09:03so that you're sure that when you present it, you're presenting it correctly.
09:07So, now I've actually saved the most burning question in everyone's mind.
09:12So, one point often brought up when AI is involved is the fear that it would eventually replace humans in the workplace
09:19or replace us from our jobs.
09:21So, Yves, is that true that AI will replace humans in the workforce?
09:26But the answer to this hot topic and more about AI will be discussed in the next part of this episode.
09:32See you there!
09:33We also sat down with Sir J. Mark Tordesilla.
09:36You know, generating summaries of more complex documents or even transcription of information.
09:51Transcription by ESO. Translation by —
10:21Transcription by ESO. Translation by —