Lewis – Google – The Future of NLP.mp4

FESCH.TV INFORMIERT:

Introduction:
med Zitouni is the Director of Engineering at Google and in our latest episode, Imed speaks to Lewis Adams-Dunstan about building Elite NLP Teams.

Imed has a Ph.D. in Computer Science and in our LATEST PODCAST EPISODE he draws on the experience gained throughout his extremely impressive career, working with some of the world’s most prestigious businesses, to discuss the following:

✅ His journey through tech to NLP
✅ The evolution of NLP; human and computational understanding
✅ Driving innovation at a leading tech business.
✅ The future of NLP.
✅ Advice for NLP as tools for business.

Here is just a snippet but you don’t want to miss the full podcast episode!

#google #engineering #nlp #future #computerscience #computer #technology

Transcription:
Lewis Adams
I want to kind of just press on to the next or at least the kind of topic we were aiming to discuss. And that was what does the future look like for NLP? So you’ve got a slide here. We talked about the kind of virtual assistance and where we are up to date, but what’s next from your perspective?

Imed Zitouni
We just touched a little bit on a few of these things when We talked about what is coming, which is we are calling the hang low about the US and all of that. So I see that we just scratched the surface on this area, and there are many to come. And while I’m saying this, it’s not because we have GPT-3 today. Really. I believe GPT-3 is great. They are meant to memorize, but there is plenty of things missing around reasoning.

You talked about emotion, you talked about behavior. So all this area, I see them coming now with more computation, with more data, with technology moving forward, technology, I believe, DC or GPT-3 and similar technology, they will be available for everyone at one point in the future, maybe in eight years, maybe in thirty years. It will happen. now when you reach that. So we are in a position where we memorize well, the language. And now if you look to small kids, usually that’s what happens, right? You memorize, you memorize, and then you start to think about how to reason about what you memorize.

And we are missing that today. We are missing that even in technology like the GPD-3. But I believe that the future will allow us to open up in terms of reasoning if it opens up in terms of reasoning. There are many great things where I see the applied, whether that for health care, whether that for this multimodal capability where you have the language, mix it with a gesture, mix it with the virtual reality, a lot of opportunity on this space. You can travel without necessary traveling.

You can have the experience and connect with others, and everyone is his place. We can advance the self-driving cars, all of these. I am a big believer that will make advances in those directions. And we’ll get there. plenty of other opportunities, a few things that we should pay attention to that as we are moving forward as well, which is the trust and the fairness in AI. And those areas that we need to pay attention to. There are buys in the data today, and accordingly, our model somehow gets bias quickly.

There is a bias in the representation. There is a bias in historical data. There is a bias in measurement, the temporal, the quantity production, the sampling, the population bias, the aggregation bias, the funding bias as well. With the popularity bias, there is a lot of buys that are affecting the models that we need to address. So we do have challenges, but also I see where the challenge is. A lot of opportunities come.

Lewis Adams
There’s a business that I’m supporting that’s doing something pretty exciting as a gamer. It’s exciting for me, at least, but they’re creating immersive gaming, whereby they’re introducing NLP to allow you when you’re using AR VR to have interaction with the characters on their game. In that, for me, is excitement. I’m not sure if your girlfriend is going to be super excited. I’m going to spend even more time I’m playing games.

Imed Zitouni
Alright, so that’s why in this slide, you have the avatar part of it, which is referring to what you just said.