Solutions to Drive Safety and Privacy in the Metaverse with Tiffany Xingyu Wang - Step into the Metaverse podcast: EP33

Good day everyone and welcome to the Step into the Metaverse podcast. My name is Marc Van Reimenam and I'm also known as the Digital Speaker. I'm a future tech strategist and I think about how emerging technologies change organizations and society. This podcast is part of my new book called Step into the how the Immersive Internet Will Unlock a Trillion Dollar Social Economy. The book is available as an E book, hard copy and audiobook. As part of this book I did nearly 100 in depth interviews with the stakeholders involved in creating the open Metaverse, the investors, the developers, the artists, the collectors, etc. And I've turned a selection of these interviews into this podcast as I think that these conversations are too valuable to leave them on my computer. In this interview I interviewed Tiffany Xinju. Tiffany is the Chief Strategy and Marketing Officer at Spectrum Labs, a venture backed AI company that keeps over 1 billion users safe online. She was named as Forbes most Entrepreneurial CMO and the most Innovative CMO by the CMO Club. She's also co founded Oasis Consortium, a think tank to advance digital sustainability through ethical standards and technologies for Web3, which launched the first ever user safety standards. During the interview we discussed sustainability in the Metaverse, the dangers of the Metaverse, and the need for ethical standards in the immersive Internet. We also covered the Oasis Consortium that she founded and its important role in creating an open metaverse. Enjoy the conversation.

If you think about the way I look at the world of trust and safety for the Metaverse, of course you need the governance layer, the guidepost, which basically the standards are doing and it's very important we can get back to it. Why? And I think for me it was the reason why we ended up where we are today in the web 2. You have 40% of US Internet users reported they were subject to harassment or hate speech. Every 39 seconds there is a data breach and then I think you can arguably say AI discriminates human beings because for example facial recognition engine recognizes white skinned males 34% better than dark skinned females in certain circumstances. Right. So all those issues were fundamentally due to the fact that we didn't have safety by design, privacy by design, and inclusion by design embedded at the beginning of this iteration. So as we have this opportunity moving into new chapter, it is moment in time to take all the lessons we learned from the social web into the new path. And then the standards basically summarizes all the lessons we learned and created the best practices that any new platform can jump start with and all the existing Platforms can literally use it benchmark against what they're doing now and identify how to improve things they are doing. So you have this whole component of governance and then another piece which you mentioned already in the book is really the workforce. Right, Great. You have already the roadmap who going to do it? Well, the issue is that it's not a subject which is taught currently in college. There's always who don't do it. It's literally the biggest question I got from everybody I work with in the space. People ask me literally every week where can I find that person I can hire. Right. So there is a big shortage and demand for the trust and safety officers and all the professionals under that umbrella. So that's why the concept I propose is you need to think about this role from leadership perspective. The good analog was that not until the chief diversity officer is a role was created did the web really invest money and resources through corporations into diversity, equity and inclusion because you don't have somebody to own it. So it's very important to have that leadership vision and then drive a whole workforce toward it. And then there is a LinkedIn learning course that LinkedIn going to publish February 14th. I think it's a very interesting day. They pick. So I basically walk people through what kind of talents you need and what kind of knowledge you need. And if somebody who take the role and enter into a company to do trust and safety, how you get a lay of hand of what's going on and make change. So there is this workforce component and the last component I would say is the technology. So if you think about it, the technology brings us this fascinating world we live in today. And because safety was an afterthought in Web2 we end where we are today. But you can also think about how technology can also be for good, right? Like what kind of trust and safety technology you can adopt. As we open this new chapter of the metaverse we can embed from the very beginning. Why is it important? Because most of for example human moderate. Most of the content moderation in Web2 started with human moderation. Then the keyword I'm more happy to talk more about if you're interested. But the second is keyword based moderation and then AI moderation. The human moderation is very traumatizing for human beings. You can read so much. Yeah, yeah, right. Horrible. Yeah. And accuracy is about 70%. The reason where there is a job is often because those human Moderators work on 8 hour shift exposed to traumatizing content. At a point people just Hit wall, right? And also because the cultural differences, right? A human moderator based in Philippines might not be very in tune with American politics and vice versa, right? So there is this error in it. Keywords based moderation accuracy is about 50% is like flipping a coin. The best example that I give is like when I'm on the live show, I usually joke about they will blip me, right? Because it's best. But the word F, right, Like it is profanity on its own, right? So if a machine look F, they will say, oh, this sentence is profanity. But what if this F word is in the context of this is F awesome? There's nothing wrong about it. You see that everywhere. So you cannot really flag that out to human moderators of platforms to take action on somebody who say it is F awesome. But if this is F awesome is in the context of child pornography, human trafficking, which are simply illegal. If that happens on your platform, that is a highly severe situation. So, so how would AI understand that? Well, you need to look at the context, right? You need to see what sentences go before what is the time of the day and what is the response of the person who received that comment. So contextual AI with right data can bring that to 90%. So the question like how when you build a new platform as a Metaverse builder to embed that from the very beginning, because it is like a snowball. If you don't do that at the beginning, then more toxic content, more people think it is allowed and the more toxic content and then before you know it, your brand is totally damaged and your users are groomed into the toxic way of acting. And then in Metaverse is way more way trickier because the whole idea of being immersive can amplify the impact, right? If somebody grope you like assault you in text world, it's okay, but not. It's not okay, but it is harmful. But if somebody is in your avatar groping you and the identity you put your spirit in and it's very impactful and the persistence increase the velocity of how fast it can get to you. And then all the fantasies or the realization fantasy increases the exposure. So if you don't embed technology at the very beginning, arguably the metaverse cannot sustain because people will not mentioning the risk of certain governments by stepping right. Like, for example Australia. Take example takes away less permissive than the US based on the web 2, right. Australia has a lot of measures in place for online safety. So technology is very important to consider. And then it's very interesting to also talk about, if you want, is the evolution of not only technologies for good, but like how you evolve that technology for good. For example, if you go into text, moderation was already hard, like the examples I give you. What if it's voice moderation, right? When you have multiplayer play the game and how would you know which sound is from whom very quickly in real time, and knowing how expensive it is to process the audio data, right, it's super expensive. And then the technology is not necessarily there because most people today use transcription. They take your words on the paper, they applied AI on text, but there's so many intonations, little words that potentially can signify it is okay or it's not okay. And the technology is only developing, not mentioning if it is in the whole digital identity. What if it is the touch, what if it's the punch? And how you embed that trust and safety concepts and therefore workflow and therefore into the code as you develop the engine for the metaverse is a whole new world, like I'm personally super excited about, because if you want to get it right, it's just not lip service. It's intellectually interesting how the people, when they code, they think about it. And those people will literally decide how this new era of digital society will behave. And I don't think people realize the impact as yet. And I think it's really worth talking about at this point.

Yeah, I completely agree with you. I think there's that. I've identified quite a long list of dangers of the metaverse, and they're quite substantial. And I think the metaverse is fascinating, but it's definitely not without its challenges. So I'm curious, how did you get started with the OASIS consortium and what's the direction you want to take it to?

Yeah, so it got started when I realized there was this big wide space of opportunity for the whole web to potentially get it right. So I moved to US about six years ago. And when I first came here, I told you like before, I was clean energy investor for 10 years, right? And like when you work in clean energy, you would know that 15 years ago, the mainstream is coal and oil and gas. The metal is mining. You tell somebody that clean energy on the default, you will be locked out of ass. People will say that's just 5% of the whole consumption and the production. And it's too expensive. There's no greater parity. And just look at where we are today. All the lead buildings, like our EED buildings, all the green energy, all the new movements. So Sustainability in our physical world is the the question that worth asking is what if trust and safety in Web3 is not an afterthought? What if it will become a de facto if we do that right. So there is that component kind of push me to think about how do we shape this digital sustainability? Right. The mission of Oasis consortium is what we say is to drive digital sustainability through ethical standards and technologies. So that's the mission and that was the background which I saw what happened in physical world I experienced and I felt like I didn't do enough 15 years ago to advocate for that. And this time I should. And another reason is so when I moved here I first work I was leading the AI product management for data management platform and Salesforce. And I have patents in marketing artificial intelligence engines. So I know exactly the power of having how much data data and then make predictions. Right. And that leads me working with a lot of advertisers and brands. Then I joined the company started with the same founders who created companies acquired by Salesforce. And then the focus is actually on trust and safety. So the idea is to use AI to to drive trust and safety and car moderation. And I work with so many platforms across dating, gaming, social media, e commerce. And I realized that there is a big gap between the world I know through Salesforce and the world I know through Spectrum Labs with content moderation companies. And then the gap is that the brands and advertisers talk about brand safety. So they say I don't want actually my brand to be adjacent to toxic content. So they led agencies to actually implement something. Right. They didn't describe this. I just don't want the results. But there isn't a pass. So you have one side on that, you have the other side of the people who spend millions of dollars to in content moderation to make a change. But there isn't an agreement between the two, right? Oh, I know actually people from both sides. Is there anything I can do? And also realize the biggest issue interest and safety is not technology. Even though I mentioned it's very hard, there's nobody who agree to standards. Right. Two dating platforms can have different definitions when it comes to sexual harassment and not mentioning a gaming platform. And dating platforms can disagree to what is considered acceptable. So that was the origin for me saying I have the personal networks and I have the knowledge and what if I do something about it? And then we started that was August 2020 after Covid first happened. And then again brand agencies the client development office of Dentu. Right. One of the biggest agencies, I get a couple of kind of trust and safety leaders in dating like the ME Group, Grindr, Riot Games, Roblox. So we all came together, say what we can do. So we shaped the safety board of I think about 20 individuals from different platforms, agencies, other nonprofits and government agencies. And then with the goal to launch the standards this year. And then we did it. So we basically compare all the best practices we can find across different platforms. We consolidated all the pitfalls and the lessons we learned from the web. Two, we compare some inconsistencies and gaps and then we push forward the standards, which we believe that the new platforms can really just take it and then use that as a checklist of what they should do and think about because you have a costar problem like any start, any platform want to do anything else, where are you going to start with? And then existing forms that allows you to say, hey, those are the things. And the best practices should look like what I did, what I didn't do and then improve it in a way. The industry has already done that, but in a very fragmented way. And all what we do is just to synthesizing it and catalyzing it and with right partners to push that forward.

Yeah. So, and what is now that you have the standards completed, what is the next step of the OASIS consortium then?

Yeah, so there are two things. One is, gosh, it's amazing that from where people didn't even realize there's a problem to now there is a past and people want to have that right away. The truth is Rome is not built within a day. The first thing that needs to happen is tens and hundreds, if not thousands of platforms start to use it. So the first thing I'm doing now is we see overwhelmingly actually positive feedback and action. So I think within the first day we launch first week with 200 downloads of the standards. And then there are different media help us even further scale it. But I think the first thing is we want to ensure it can be used. So we also partner that we have announced it. So I'm basically working with a couple other nonprofits to launch a partner ecosystem. So think about like six to eight other nonprofits who have been there a long time in brand safety, in ad frauds, certifications, potential in human trafficking right at the counter, human trafficking in trust and safety, education, in gaming best practices and also the safety tech network, for example in UK so government agencies. So we try to launch a panel of people where the standards can help their work furthered and further distributed. So I Think the first step is now we have the awareness and we have these standards we need to drive the adoption and we actively just track how many people download and use it. And then from there it is in active conversation. But I couldn't give more information because it's just like in terms of the how the system going to figure it out. But you know, off the record with you is like my question is like do I want to build an audit a certification system within Oasis or I should partner with an organization who is specializing certification and make that available to different platforms? I think less important about how it looks like. I think the key message here, maybe for the book and for you, is that when enough is enough, right? So we started where it was such an afterthought. Now it becomes something people can use right away to what I say called safety by design. And then existing platforms can use that for self assessment. So it's already major leap forward. Do I want to spend more energy to horizontally scale that like more people use it and more people use that for self assessment or I want to focus more my time to say let's build a certification to audit limited because when you do that you only, you only have that many resources every single point in time and only focus on the big platforms and let them to do better student audit system which has more impact. And I personally in the camp the first is saying that because we open up a new metaverse, let's help those people who didn't have the legacy of issues and get it right from the beginning. So adoption awareness and self assessment is where I focus more and hopefully with the right partners we can turn the standards into a certification program that anybody can go through certify and put that name out there. Right?

Yeah, I think that that's very, very important. And one of the things that I write in the book, and I've actually been telling that for a very long time, is the idea that what we would need also related to AI. So you basically want the financial auditors to shift gear and to become AI auditors and that they are responsible for the orders that they're doing. So that we know that the AI does what it says the organization says it does and that we know that it's not the black box and that it's not biased. And I think the same would apply for the metaverse that we can make sure that those platforms who have millions of users that they adhere to your standards, but also other deers around safety and bias and ethical use of AI for example. So I Completely agree with you that I think that that's very, very important.

Yeah. And to that point, so I didn't mention, we will launch a privacy board. So you cannot talk about safety without privacy because the way for you to keep people safe is you actually have the data to actually eliminate toxic behaviors. You need to be very, very careful with privacy. Right. You cannot just collect data as you want. Right. There are regulations like gdpr, ccpa, there's a Virginia, US Virginia version of that to protect that. So how will you be able to do safety while being compliant with privacy is not talked often like people don't actually know. For example, in those regulations such as gdpr, it is carved out saying that trust and safety cases actually can process the data when it is to save one's life, when it's to actually being compliant with legal obligation, etc. Right. So all those nuances is something I personally care a lot about. Like you cannot talk about safety without privacy.

I completely agree. Because I think the victim of trust is privacy.

Exactly.

I've been actually thinking a lot about this for the past five or seven years with my different companies I've had. And I think that the right way to go here, and it's also what I'm writing in the book is a concept which I call anonymous accountability where you can remain anonymous but you are being held accountable for your actions.

I love it. I really love that. Yeah.

And nowadays it's finally possible with blockchain technologies with zero knowledge proofs, with self sufficient identity that one, we can verify that you are human and two, we can verify that you are a trustworthy human without knowing who you are. And I think that that is absolutely key. And yeah, from that perspective I think it's great that you focus on building, launching a privacy board as well.

Yeah. So I think that's something new. Ultimately it's just drive more safety awareness adoption and then launch a privacy board. But it really comes back to at the end of the day, the initial thing I said is how you build digital sustainability or how you build a sustainable metaverse through those ethical standards and also through that movement to drive more ethical enterprises and technology companies. Right. You know, you need a safety, privacy and hr. Right. Like the inclusion technology companies to, to support those enterprises who want to do that things ethically. And then that's one thing like how you help those companies. So we are thinking of, we haven't announced it. I would try because it's harder because I wouldn't say it's easier like to a certain degree it's easier to find those big logos and big platforms, let them to sign up. It's actually very hard. But if you think about there is a benchmark, right? There is best standards, all we need to do is to make sure they can be adopted by more people in a very actionable and specific way. That's what we did. But how you actually be able to select the ethical technology companies? For example, like today, I talk with a Metaverse company. What they do is the touch. So how you create in the metaverse, you can touch somebody, right? Like that sensory. And then of course, they think about trust and safety and ethical stuff. How would I be able to know for sure the trust and safety was down in the technology? Right. Because there's a lot of transparency needed. So something we didn't talk about publicly enough, but I think curates a generation of innovators and entrepreneurs who want to develop ethical technologies or who want to bake ethics at a design phase for whatever they build is a goal that we have through the transcendent body of Oasis. And then last thing I think is the workforce. Right? Like standards give people basically a tool, a playbook to get started very quickly. But I think there might be a lot of opportunity to partner with college and academia to make sure things can be taught in school. And LinkedIn is a way, of course, I find, you know, in the modern days, you add your Kusora, LinkedIn, Udemy, all those platforms can be also good partners to do that.

Yeah. Okay. Well, thank you so much. That's very interesting and very useful. One final question I have for you and a question I ask everyone what we think is the future of the Metaverse, which is quite funny to ask because we even haven't even built the Metaverse. But if you had to look five, 10 years out, what are your ideas?

I don't know what it exactly it will be. It might even not be called Metaverse. But what I know is it should not be what I call a mirror. First, for me, Metaverse is an opportunity, is an opportunity of a new universe where we can carry on the things we like about what we have built and amplify that into that new universe and get rid of the things we didn't like or the mistakes we made. The example is, for example, Fortnite has Ariana Grande concert. And in real time, over millions of people gather together and socialize with each other that we like. We like being able to socialize with each other and meeting new people and defying the physical gravity. Right. We like that amplify it. We didn't like being harassed, we didn't like being trolled, we didn't like that our data monetized without us knowing it, without us being compensated. Well, then fix it. So I think Metaverse, what I know for me is an opportunity to open a chapter, to carry and amplify the things we like, mainly through the new technologies we build. Because all those Metaverse, like all the game engine to render the real physics, you can see the digital self. Right. And all the multi sensory technologies, all the payments that enable creators to be diverse. Right, the creator economy, all those things are great and we should amplify them. What we don't like is we don't like safety, we don't like privacy, we don't like inclusion issues. And that's the opportunity for us to fix it.

Yeah, I like that. That's a very good finishing point. So thank you so much, Tiffany. I really appreciate your time and input.

Yeah, thank you.

And that wraps up the conversation with Tiffany Xinju. If there's one thing that stands out for me from the conversation, it is how we can use AI to drive safety in the metaverse. But achieving this is a long journey that requires careful thinking. Which is why in my new book, Step into the Metaverse, I offer three additional solutions to drive safety in the metaverse. If you enjoyed this conversation, don't forget to hit the like and subscribe button on your favorite podcast platform, so you will be notified when any of the other interviews will become available. Thank you and see you in the Metaverse.