69 minutes
SDS 409: Succeeding & Networking In the Virtual Space
Subscribe on Website, Apple Podcasts, Spotify, Stitcher Radio or TuneIn
Very inspiring guest from ACS. This will be particularly helpful to new and intermediate practitioner. We covered AI startups, hackathons, contemporary AI questions and topics, NLP and computer vision, the AI bubble, and a lot more.
About Steve Nouri
Steve Nouri is a data science leader who has evolved the way people look at AI and innovations. An entrepreneur, investor, author, academic and technical manager by profession, he aims to inspire people through the latest technology trends and projects and empowers prospective data scientists through high-quality education and learning materials.
He is a member of the Forbes Technology Council, an alumnus of MIT and UTS, ICT Professional of the Year Gold Disruptor, and an accomplished influencer with over 200,000 followers on LinkedIn. From humble beginnings as a software engineer and startup founder, he has worked in different IT roles before joining the Australian Computer Society as the Head of Data Science and AI, and as an AI evangelist has spoken at countless international conferences run by IBM, JP Morgan, CSIRO and AWS.
He currently lives in Sydney, Australia, and spends his time sharing technical knowledge with prospective students and advising public policy towards a more sustainable and beneficial understanding of AI and deep tech.
Overview
ACS, the Australian Computer Society, oversees all the computer courses in the country and assists IT professionals in success. At ACS, Steve wears two hats. He is head of data science as well as head of the community development work they do. He found himself in this line of work after spending work as an advocate for the benefits of data science. In their work through River City Labs, they help startups cultivate their culture and customer service features.
Steve also has a big passion for hackathons, he’s participated and helped run several. Steve started mentoring hackers for hackathons and then, upon joining ACS, he took on the role of running hackathons. Their Flatten the Curve hackathon was a 9-day hackathon featuring over 2,000 people working to find solutions in education and business during the pandemic. They’ve hosted several others across Australia, all of them recently have been virtual which involves added time for virtual mingling and networking through cultivated Slack channels to find needs for each team.
We shifted gears to discuss Steve’s online workshops where we discussed some of his work in helping startups and the common issues he’s found. Interesting, we discussed the AI bubble that’s been created from hyper-inflated expectations and how this affects the trust the public will have in startups and AI in general. A way to combat some of these issues is showcasing real AI and its measurable impact, rather than the businesses that claim the use of AI for publicity. We also talked about the importance of strategic data acquisition and the nuances of explainable AI.
We shifted into the final topic: personal branding and the importance of it. He believes that establishing a personal brand over digital channels is incredibly important to thrive in today’s environment. Attracting recruiters or business partners happens online, over social media, in virtual courses, and through other non-physical means. And Steve will be offering a number of workshops and resources on this topic later this fall. For those who are timid about stepping out into the digital space as a contributor, it can be as easy as sharing a book recommendation. You don’t need to be an expert to offer something of value to someone.
- Steve’s work in the Australian Computer Society [4:32]
- River City Labs [12:22]
- Hackathons during the pandemic [16:21]
- Choosing a path in AI [26:09]
- The AI bubble and its implications [31:09]
- Strategic data acquisition [38:04]
- Explainable AI [43:50]
- Creating a personal brand [51:35]
Items mentioned in this podcast:
- River City Labs
- HackMakers
- Machine Learning Engineering by Andriy Burkov
Follow Steve:
Episode Transcript
Podcast Transcript
Kirill Eremenko: 00:00:00
This is episode number 409 with Head of Data Science and Artificial Intelligence at the Australian Computer Society, Steve Nouri.
Kirill Eremenko: 00:00:12
Welcome to the SuperDataScience podcast. My name is Kirill Eremenko, Data Science Coach and Lifestyle Entrepreneur. And each week, we bring you inspiring people and ideas to help you build your successful career in data science. Thanks for being here today, and now let's make the complex simple.
Kirill Eremenko: 00:00:44
Welcome back to the SuperDataScience podcast everybody, super excited to have you back here on the show. Today, we've got a very inspiring guest calling in from Sydney, Australia, Steve Nouri. Steve is the head of data science and AI for ACS or the Australian computer society. This is an organization that oversees a lot of things that are to do with technology in Australia, and Steve has a lot of experience in this space and came today to share his insights and some inspiring thoughts. So this podcast is going to be extremely valuable to you if you're just starting out into data science, or if you're an intermediate practitioner in this space, looking to grow your expertise further. And here are some topics that we are going to be discussing today. So we cover off three main parts in this podcast. The first part is about inspiring applications of artificial intelligence, what they're doing at ACS and how they're supporting communities. How you can get involved with initiatives like this, whether you're in Australia or outside of Australia, because these initiatives exist in other countries as well.
Kirill Eremenko: 00:01:58
And also the ACS runs some initiatives that are global. For instance, they're doing hackathons virtual hackathons, which you can participate in. You'll learn about AI startups, hackathons, what it's like to be running a virtual hackathon, and how they happen. So if you ever wanted to find out about hackathons, this is a great place to learn. Steve himself has participated in over 10 hackathons and he's run several of them. So he's very knowledgeable in this space. Then in part two of the podcast, we're going to talk about some contemporary AI questions. For instance, what's the best place to get started in artificial intelligence? You'll learn why NLP and computer vision are Steve's go-to choices. We'll talk about the AI bubble, and what the implications of these hyper inflated expectations of artificial intelligence are, specifically for artificial intelligence startups. You also will find out what you can do to help remove this bubble or combat this bubble. We'll talk about strategic data acquisition and why it's such an important step for any company in the AI space. And we'll talk about AI, explainability and where the world is going with this topic.
Kirill Eremenko: 00:03:16
And finally, in the third part of this podcast, we'll talk about something that's very dear to Steve's heart, and that is building a personal brand in the space of AI. So you'll get some valuable tips from Steve who himself has over 200,000 followers on LinkedIn. So he definitely knows what he's talking about when it comes to building a personal brand. You'll find out why it's absolutely important for everybody in this space to be doing that, and how this can make our community, our AI and data science community, even better. So we've got a very exciting podcast with lots of topics coming up, can't wait for you to check it out. And without further ado, I bring to you, Steve Nouri, Head of Data Science and Artificial Intelligence at the Australian Computer Society.
Kirill Eremenko: 00:04:05
Welcome everybody to the SuperDataScience podcast. It's super fun to have you back on the show. And today we've got a very special guest calling in from Sydney, Australia, Steve Nouri. Steve, how are you going?
Steve Nouri: 00:04:16
Hi Kirill, thank you for having me.
Kirill Eremenko: 00:04:18
Super excited, super excited. Already so pumped after our chat before the podcast, super pumped. There's going to be a lot of exciting topics to talk about. You're excited as well?
Steve Nouri: 00:04:28
Very excited. Yeah.
Kirill Eremenko: 00:04:31
Awesome. Awesome. Well, to get things going, tell us a bit about yourself, please. You are at the ACS, right? So the head of data science and AI at the Australian computer society. Can you please tell us, what is the Australian computer society and what is your role there?
Steve Nouri: 00:04:48
Yes, as you just said, I'm head of data science and AI at Australian Computer Society. First of all, ACS is peak industry body for IT and ICT professionals in Australia. It looks over many different activities, accrediting courses in Australia, all the tertiary courses, computer courses in Australia are accredited by ACS. And also we help IT professionals to be more successful in their career life, running many events, workshops, and making more opportunities for them to connect with leaders and also running some courses, even [inaudible 00:05:41] certifications for their accomplishments. So these are all the different things that ACS does, and many other stuff, like for immigration, the assessments for immigrations. And so it's a large organization helping Australian IT society. And my role is actually an interesting one.
Steve Nouri: 00:06:06
So I wear two hats at ACS. First of all, it's head of data science and AI, and it looks after the data science function, helping other functions to do data governance and data analytics. And that's the normal data science job. The most interesting part, which I really love as much or maybe even more, is I'm also responsible for AI community development in Australia. It's a kind of an advantage of this role. So I get to go around Australia, talk to people, talk to enthusiasts and professionals, helping them to understand more about data science and AI, and how they can upskill, reskill, or learn it quicker and sometimes deliver keynote speeches, panels, at ACS. And yeah, that's pretty much very interesting to be connected and directly talking with professionals and leaders in that area.
Kirill Eremenko: 00:07:22
That's so cool. How did you get into this line of work in the first place? It's such an interesting role to be in.
Steve Nouri: 00:07:28
Yeah, so it kind of naturally happened because I was always an advocate and I was always an ambassador for AI, for good data science, for running different events in my personal time. Running meetups. And also I'm super proactive on social media, specifically LinkedIn, and it was a complimentary of my technical understanding and what I used to do as a data scientist and as an educator so meshed together and blended nicely, and when I joined, they gave me all these paths together as a whole function and I'm loving it.
Kirill Eremenko: 00:08:29
And I love how proactive you are on social media, because that's actually how we met. You shared this cool video of holographic avatars where people are dancing and then they're replaced, or actually augmented, with holographic avatars. I think that was really cool. I was very impressed by the video, and I'm glad that you share around these things, and no wonder you... I think you have... was it 200,000 followers on LinkedIn? That's a huge number.
Steve Nouri: 00:08:59
Yes. 220 something right now. Yeah, that particular clip was actually very interesting. I share these inspiring, interesting innovations all the time. Daily, I will share two or three of them, and they are mostly around AI, artificial intelligence, data science, and augmented reality, things that are kind of high-tech and very interesting right now to help people to know more about the applications and also to get more enthusiastic and inspired and excited about it. I have received many messages that tell me that we changed the course ... literally, there was a guy saying that I was doing mechanical engineering, which is a great topic and we definitely need a lot of people, but he just changed it to artificial intelligence just because of those interesting posts that I was doing.
Kirill Eremenko: 00:10:07
Oh, wow.
Steve Nouri: 00:10:07
And he just said, "I was inspired by that. So now I changed it to AI," and I said, "That's great. Hopefully you're going to like it."
Kirill Eremenko: 00:10:14
That's so cool.
Steve Nouri: 00:10:15
And that articular post went super crazy viral and right now it has around 3 million views.
Kirill Eremenko: 00:10:20
Wow.
Steve Nouri: 00:10:20
Which is crazy.
Kirill Eremenko: 00:10:23
That's amazing.
Steve Nouri: 00:10:23
I wasn't expecting that.
Kirill Eremenko: 00:10:25
That's amazing. Amazing. Hey, everybody hope you're enjoying this amazing episode, and we've got a quick announcement and we'll get straight back to it. The announcement is that DataScienceGO Virtual number two is in town. It's happening on October 24th-25th this year, and you can get your tickets today at datasciencego.com/virtual, and the best part is absolutely free. We've got some amazing speakers, amazing workshops for you to attend, and of course the super cool part is that we've got networking.
Kirill Eremenko: 00:10:58
There will be several 3 minutes speed networking sessions, where for 3 minutes you connect with a random data scientist from another part of the world, or maybe from your part of the world. You get to chat for three minutes. If you like the chat, if you want to connect, you hit the connect button, you stay in touch. This was by far one of the top features of DataScienceGO Virtual number one. So many people got such great connections, stayed in touch, and some crazy stories came out of that. So we're going to repeat it, and we want you to connect with your fellow data scientists. Once again, it's absolutely free. Register for your ticket today at datasciencego.com/virtual, and I'll see you there, and now let's get back to this episode.
Kirill Eremenko: 00:11:39
And we'll talk more about personal branding and becoming an influencer and why it's important to build a brand, I think towards the end of the podcast, but for now I wanted to start on saying, as we discussed before the podcast, it's a coincidence... I just realized this, that SuperDataScience and DataScienceGO, our businesses are registered in Australia, are Australian businesses, and I'm speaking with the head of AI and data science at the Australian Computer Society. I wanted to commend you guys for doing a fantastic job for River City Labs, what we spoke about before, a place... it's like a co-working space, but more than that. Can you tell us a bit about that? Because I find it very inspiring and I know, not just in Australia there are co-working spaces like that, they are all around the world, and I think people who are running AI startups or who want to get into the space, want to progress in the space of AI, will benefit a lot if they go to a place like River City Labs. So if you don't mind telling us, what is that all about?
Steve Nouri: 00:12:45
Yes, you're right. So River City Labs is part of ACS and part of the labs in general. We have three labs right now in Australia, in Sydney, Melbourne, and Brisbane is called River City Labs, and the labs are not only coworking space, they also help startups and scale-ups to understand how to get to customers, how to grow, and we help them to scale up and become more successful and then graduate from our labs and just become a corporate, and we had a couple of really good examples, really good examples that were graduated recently. Fathom AI was one of them, and-
Kirill Eremenko: 00:13:42
I've heard of them. Fathom is a reporting tool.
Steve Nouri: 00:13:45
Yes.
Kirill Eremenko: 00:13:46
No way, they came out of River City Labs?
Steve Nouri: 00:13:48
One of the labs, the one in Sydney, yes.
Kirill Eremenko: 00:13:48
One of the labs.
Steve Nouri: 00:13:48
Yes.
Kirill Eremenko: 00:13:50
That's so cool. We use Fathom for reporting. It's a very affordable reporting tool. If you hook it up with Zero, which is a New Zealand company, for accounting, and you get all these amazing reports. I didn't know that. That's so cool.
Steve Nouri: 00:14:04
Yes, they do a lot of good reporting and stuff, and also there are other AI and high-tech startups that are working in the same path.
Kirill Eremenko: 00:14:16
That's awesome.
Steve Nouri: 00:14:17
Yeah, it's a place that... we want to help the ecosystem, make the ecosystem so you're not only helping graduates and beginners. We also are helping industry startups and academics, and kind of connecting all together to make sure we, as an industry body, we're helping make the whole ecosystem work nicely together and help each other grow.
Kirill Eremenko: 00:14:47
That's so cool. I attended one of the talks at the River City Labs. It wasn't even a talk, because I had heard of this thing for a few... maybe even years, and I was putting it off, but then I attended at the start of this year, before coronavirus. I think it was in February. It's once a month, I think is the last Friday of every month there's a pitch night, where five or six entrepreneurs in the space of technology get five minutes each to pitch their idea to an audience, and there's a panel of judges, and then they're given some feedback and they're told how they went. So it was a kind of opportunity to practice pitches, and maybe you'll find an investor, if one happens to be there. I enjoyed it thoroughly. There was some interesting, diverse ideas. I remember one guy was pitching an idea about using AI to help with robotic limbs for people who lost an arm or a leg, making them more advanced so that they can make more advanced movements and powering out of AI. So yeah, ideas like that or nights like that, fantastic, I think.
Steve Nouri: 00:15:53
Of course. Yeah. No, that's great. That's great. So there are lots of initiatives. They run a lot of initiatives, mostly they help, connecting these startups and scale-ups to customers and also to investors. That's very important for them at the earliest stage to become successful.
Kirill Eremenko: 00:16:19
Absolutely. Let's talk a bit about hackathons. You are an avid fan of hackathons, as you told me, and more, where you run hackathons. You run the biggest hackathon in Australia, and you put it together in record time. Tell us that story.
Steve Nouri: 00:16:36
Yeah, you're right. That's a passion of mine. I was always involved with hackathons. Six years ago, I participated in the first hackathon, and from then, I just loved it. I remember very practically searching for hackathons on the internet and trying to find something for the next weekend. I was just so addicted to hackathons that I couldn't just be without any hackathon. I was just like, "If there is any geek people doing some technical stuff together, I'm going to be there." I participated in more than 12 hackathons before I got too old for hackathons.
Kirill Eremenko: 00:17:18
Too old for a hackathon.
Steve Nouri: 00:17:23
There's no such thing as being too old for hackathons, let me tell you. It was a joke. Because a lot of people ask me is hackathon for graduates or beginners. No, there's not such a thing. Hackathons are for people who are enthusiastic about tech. They want to do something interesting in a short amount of time. They want to have an impact, and also they want to learn from each other. Many good, interesting ideas come from hackathons. I have stories about that.
Steve Nouri: 00:17:51
But, first of all, when I did a lot of hackathons, then I was a little bit more senior. I wanted to step up into running hackathons and mentoring hackers. I started mentoring hackers in hackathons, but then when I joined ACS, there was a nice synergy between ACS as an IT community and running hackathons for IT communities.
Steve Nouri: 00:18:30
We did our first hackathon, Flatten the Curve Hack, during COVID. We managed the whole thing, my team, and also with the help from other teams across ACS. The marketing team and membership team, they all came together very nicely during nine days. We literally had nine days to pull it off. At the end of the day, we managed to get around 2,700 people across Australia and New Zealand to work together, come up with solutions for problems that we have in the pandemic. Education, businesses, everybody was impacted by COVID. We were still in the same situation, but back then, we wanted to have early solutions for the problems. It's like how we can help medical centers, how we can help these businesses that has been effected and now closed down, how we can help the education centers and medical... and all of them.
Steve Nouri: 00:19:43
We managed a hackathon within nine days. They had only 48 hours to deliver the solution. On the back of the hackathon, we had very interesting ideas. Especially, we had three startups out of that particular hackathon. That's very interesting to see. Some of these people even didn't know each other. They came together during the event. They were enthusiastic about the same vision and same mission, and they started working together. Finally, they came up with a nice solution that they were happy that this will have some impact. They started a startup after the hackathon, which I'm proud that I've been involved with that.
Steve Nouri: 00:20:34
Then, from there, we managed another hackathon, the first Australian and New Zealand Defense Force Hackathon. That was also a very interesting one. We had around 1000 people involved, and people got enthusiastic about it as well. The last hackathon I was involved with, that was a global AI hackathon. We had many AI influencers as a judge and mentor there. I think we peaked with 4,200 participants. That's was super interesting. We, hopefully, each time we try to break our records.
Kirill Eremenko: 00:21:27
That sounds so-
Steve Nouri: 00:21:28
Let's see how would be the next one. The next AI hackathon would be around January next year. We're already planning for 10,000 people.
Kirill Eremenko: 00:21:41
Wow. That's really cool. I'm assuming these are running virtual. How does a virtual hackathon run? How do people get together, get to know each other? What does it look like?
Steve Nouri: 00:22:02
I think that's the most difficult part. Because as you said, just it's virtual. If they've never met each other, they need to literally meet, talk, and form a group within this 48/72 hours. I guess that's the most complex part. We try to run a lot of different things. First of all, we give them some time to mingle and talk and see if they can find people with the same mindset. Also, we make a lot of channels on the Slack that has specific reason. For example, if you're a team looking for a data scientist, this is the channel for it. If you're a team looking for a programmer, this is another channel. If you're a data scientist looking for a team, that's another channel. You can find channels that's specific for reasons. Then, it will help people.
Steve Nouri: 00:23:01
Because 4,000 people, if you just put them into place and say, "All right, guys. You figure out how to do the job," but that's not going to happen. You need to nurture them. Thanks to our mentors, we had more than 200 mentors in the Slack channel because 4,000 people needs a lot of mentors as well. They have a lot of questions, and if each one of them asked two questions, that will take ages for me, or my team, to answer, right?
Kirill Eremenko: 00:23:31
Yeah.
Steve Nouri: 00:23:32
So what we did, we also made a really nice hierarchical team of mentors. We had mentors on the ground, and we had lead mentors, helping mentors. Then, we had chief mentors, lead mentors. That was super interesting. They came up with ways that they could engage with these participants and try to make the connection happen. That's still a challenge. A lot of people will not find the team. So at the end of the day, we had to make random teams. You literally have no team and you're running out of time.
Kirill Eremenko: 00:24:20
Interesting. Where did you find the mentors?
Steve Nouri: 00:24:23
The same way we found the participants. So we put some ads out there and it's like, "If you're enthusiastic about hackathons, but you don't want to participate and you have limited time..." Because mentors wouldn't need to be available for the whole time. They could just kind of touch base with their teams once or twice during the hackathon for a couple of hours, that would be fine. But if you're wanting to participate in, you need to be available for 48 hours because the team needs you. So I guess we just made it clear that if you're happy to be part of it, but you don't have the time, then this is for you. We got more than 200 people and we selected the ones that are more senior because we wanted mentors. They need to have some knowledge and experience to do the job.
Kirill Eremenko: 00:25:17
Fantastic. Well, wow. That's very cool. I'm guessing it's not yet announced, the hackathon for January, but maybe people, if they follow you, they'll hear about it when the time is right?
Steve Nouri: 00:25:32
Of course. Yeah. As you said, it's not announced. The dates are not set yet, but they will be around January and just follow me on LinkedIn. There will be some posts coming up soon.
Kirill Eremenko: 00:25:48
Awesome. Thank you. That's really cool. All right. Well, I wanted to move on to some topics that you mentioned in your workshop because you said you do lots of workshops and try to help companies, startups, entrepreneurs and just people in the space of AI is through ACS. So I watched a few of your workshops on YouTube and I have some interesting questions I'd love to dive in. You ready for this?
Steve Nouri: 00:26:16
Right. Yes.
Kirill Eremenko: 00:26:18
All right. So in one of your workshops, you said, "If you're learning AI or data science these days, make it easy; either go down the path of NLP or natural language processing, or computer vision. These are the main growth areas." I think you said this in October last year at Macquarie University. Can you elaborate a bit on that? Is it still the case that NLP and computer vision are the main two areas, and why these areas?
Steve Nouri: 00:26:46
Yes. So first of all, if you want to just talk about what areas computer science and AI in particular will be used, the use cases are unlimited different and you cannot just categorize it in one or two. But I just wanted to help... Because there it was in Macquarie University and I wanted to help the students to figure out what is the next path? If they want to have a mindset about following some specific, let's say technology, what is it? I found out people kind of don't like theory much. They want the application. Two major applications of AI at this point are either computer vision or NLP. The main reason is because they were super complex before for any normal algorithms to handle those tasks. These cognitive tasks are super complex. So normal algorithms like handwritten algorithms, [inaudible 00:28:08] heuristics couldn't handle this. Then deep learning and machine learning in particular, they were kind of more touching on these cognitive tasks.
Steve Nouri: 00:28:24
So if you want to find some interesting use cases, I wanted to make it easy for you. These are two main use cases. Yeah, I can say that still it is relevant. As I said, the reason is obvious because we have lots more data related to these cognitive tasks. We have more videos right now. Right now, we're making another one here. It will be videos generated, sound will be generated. There will be a lot of conversation that might be transcribed, and these are really good material for the AI to digest and do something interesting. So yeah, there are many other use cases, but still, I would say if you're passionate and you don't know how to start, just grab one of these, start with it. Later, you will understand if there is anything else that you're passionate about and you can change it the direction.
Steve Nouri: 00:29:37
Don't stay still and it's like, "What should I do?" because that's another problem. I get a lot of messages that people are just like, "I'm doing research a couple of months to see what should I do." And I'm like, "If you have started with Python, you would be already a great Python programmer and you could later, figure out if you wanted to learn R then go on and learn R. It's fine." That's another mistake that we sometimes lose the opportunities by just staying still and trying to find the best. There is no such thing as the best. Dive in. You will find a way.
Kirill Eremenko: 00:30:17
Okay. I understand now, because when I was listening to the stoke, I was thinking to myself, "Well, how about other areas of technology, like reinforcement learning or sequence modeling and other interesting areas of area," but I understand why you're doing this. It's better to reduce your choice, so you don't have the choice paralysis. Start with something and then you'll be ahead of yourself in the other dimension where you're still paralyzed with choice. You'll be already ahead, and then you can choose something else if something else comes your way. So I totally get it.
Steve Nouri: 00:30:55
Exactly.
Kirill Eremenko: 00:30:56
Yeah. Okay, cool. Thank you. I think that's very powerful advice for those listening or still deciding where to start. So next one, just looking at my notes here, in that same talk, you said that more than 40% of European startups are claiming to be AI startups, but they're not. This came from some report about European startups. And you mentioned that this will create an AI bubble. What are the implications of such an AI bubble being created with these hyper-inflated expectations?
Steve Nouri: 00:31:34
Yes. Yeah. I remember Venture Capital in Europe, they've done a survey of startups and they found that 40% of them actually are not using AI, but claiming that they're an AI company. The problem with that would be a lot of investors will become more cautious about their investments. And when they hear AI, they will think maybe that's another one. And that will affect the real startups in general because now it's going to be a more complex space to be. And if you want to prove yourself, you need to do more. You need to push more to prove yourself. So it will generate some kind of negative, negativity in the space. So that's the first thing. The second one would be... The other problem is that people, in general, the audience will be a little bit skeptical about these companies, because when you hear these days, you probably have noticed these memes coming that there is a project called AI, and then you open it up, there's lots of if and elses down there.
Steve Nouri: 00:33:08
And that's a joke right now, but this is the thing that the general public are seeing as a reality, that they don't trust. So they're losing the trust, and losing the trust will make people be more skeptical. And when that happens, that will have a negative effect on the real one, which are going to have a real impact, a good impact. So that's the major implication. And also, I can then hear a lot of people saying maybe even AI is a bubble itself. Maybe there's no such a thing that is going to help. I even heard that somebody was saying that AI in general is paraphrasing programming to sell it more expensive. So you're putting another cover around it just to make it look more nice and gourmet, but it's the same thing. And these are all the negative side effects of such a problem.
Kirill Eremenko: 00:34:23
And what can either an entrepreneur who's building a true AI startup or somebody who's actually passionate about AI and wants to build a career in this space, what can they do to combat these implications of this AI bubble?
Steve Nouri: 00:34:40
So one of the things that I'm doing, as you can see on LinkedIn, I'm sharing these projects, AI projects. Every day you could see a lot of them. And that's one of the ways is by showcasing the real impact of AI, the real AI. Most of the stuff that I'm sharing are open source. So then you can actually see that this is actually AI because it has some components behind. It's open source. And if you're are really a technical person, then you can check it out and make sure that it is what it is. So I guess the communication and more training, changing the culture that people ask the question about it and have better understanding is the only way out.
Steve Nouri: 00:35:39
We need to educate everyone to know about AI. I think this is the future and what we need to do as a community. And as an ACS or me as a person who is involved, I need to educate everyone about AI. They need to understand what is called AI and how to identify the products, how to identify the impact, and what are the implications and what are the complications of AI. So this is something that I'm doing proactively on social media. And I assume that everyone who is passionate about AI should do the same to educate everyone about it.
Kirill Eremenko: 00:36:28
That's an interesting philosophy. It almost feels like collectively if we are truly excited and want this field to succeed, then the way to combat this AI bubble is to act as a filter, to take a, look out for news about AI and assess it through our knowledge and understand, "Okay, this is really AI. This is not really AI. This is somebody calling themselves AI to increase the value of their startup." And then actually just sharing the true AI. It's a collective knowledge filter of the other news that exists out there.
Steve Nouri: 00:37:11
Yeah, totally. Because this is so clear that AI has value and it will generate a lot of positive value for any businesses and for society as a whole. And to have any negative negativity around it will be so devastating for me as a person. So that's the reason that I would like to make sure that educating part is the main thing and everybody will know about it. Then if there is a bubble and if there is a startup that claiming it and then it collapses, or a project that doesn't work, it will not be seen as a AI in general, and it will be looked as one example or use case.
Kirill Eremenko: 00:38:04
Okay, gottcha. Speaking of AI startups, you mentioned that... And I think this is in parallel to what Andrew Ng was talking about, that strategic data acquisition is the core building block for creating an AI company. What is strategic data acquisition? And can you please elaborate on why it's so important?
Steve Nouri: 00:38:35
I want to just make it a little bit easier in layman terms. AI is pretty much equal to machine learning to me. I don't want to talk in academic terms. It's like, what are the differences? And what are the similarities? Let's agree that I'm not working at university anymore. I used to be a lecturer and now I'm working in industry. So I want to say that AI is pretty much machine learning. And if we agree on that, then machine learning needs data, that you cannot have any machine learning without data. We cannot run any algorithm on nothing. So then you need to capture the data in order to run your machine learning algorithm. The data acquisition is base for any AI project.
Steve Nouri: 00:39:35
And if you have your huge AI team of PhD graduates waiting in a room and you have no data, you don't have an AI. You will not have an AI until you start with the data. So if you get back to data acquisition, a lot of the companies are actually struggling to get the meaningful data that can generate some insight. Let's say if they're working with a customer, it's not easy to ask for a customer to give you all the information, and personal information and consent to use their data for some business related projects. It is something very complex. A lot of people are skeptical. I have heard that a lot of people were saying, "Why Google is looking at our search? We are not happy about it. They're investigating in our personal life." That's the reality. This is the conversation, right? I have talked about it and I have heard about it a lot. These companies need to be very strategic about how they want to collect those data, that they can also keep their customer happy.
Steve Nouri: 00:41:04
The one way that I was talking in data acquisition would be give them back something interesting, some insights that they will be happy to share the data with you. If you're collecting information about someone's weight, give them back something, for example, every month, generate a report and say, this was changing your weights. These are the other cohorts of the similar age or similar life style with you. They changed it in a better or higher or lower ways and you're getting closer to a healthy range or whatever it is. Generate some insights and don't be like a vacuum cleaner, just sucking in all these data and nothing back. People will not appreciate it, right? That's one way I would just do to make it very easy. That's one way to make them happy and engaged.
Kirill Eremenko: 00:42:19
Okay, makes sense. Also comes down to thinking about AI, not just as in, "We're going to go in and create an AI company. Boom. [inaudible 00:42:31] going to happen." Actually putting as part of your strategy or business plan, how are we going to acquire that data? As you said, there's certain considerations that need to be taken into account, like privacy. If you're in Europe, then there's the GDPR. These policies are created to help protect people's data, right? You need to think in advance, how are you going to approach that? I think your advice is very solid in terms of create a ecosystem where you are collecting data to help people, not just to use it for insights all the time to power your artificial intelligence, but create that loop where you're always giving back because if you wait too long, you're collecting data [inaudible 00:43:28] before you then create value for people, you might not have a company at all in that.
Steve Nouri: 00:43:35
Yeah, exactly. That's a great summary.
Kirill Eremenko: 00:43:40
Awesome. Okay, thank you. A strategic data acquisition should be an important part of AI startup thinking. Let's talk about explainable AI. In one of your talks, you said I understood what I just said was that sometimes AI should be explainable for sure, whether it's legislation, whether it's court proceedings or, making legal decisions, especially. It would be perfect if AI was explainable in old cases, but sometimes it's just not going to be explainable. It's really hard to explain, but may be even impossible. The question is, are we just ready to accept and move on? Accept that sometimes it won't be explainable. You said an interesting thing. You said, "I don't think we're ready and especially in Australia." Could you please elaborate on that?
Steve Nouri: 00:44:40
Yes, because explainability was a hot topic, still is a hot topic and a lot of people are talking about it. We need to make AI explainable, otherwise we shouldn't use it. I understand that to some extent, but to be clear about it, I don't want to make it so complex, but let's say AI cannot be explainable 100%. There's no such thing because of the complexity. If you think about a neural network, learning things about image, if it was so easy, we could use normal algorithms to understand it. We could use some heuristics to understand those little bits and pieces about the image. The explainability is not going to be 100%, but we can get away with some understandings. We can generate some understandings that is understandable by humans. That's there, so let's agree that 100% is not going to happen.
Steve Nouri: 00:45:52
The other problem is a lot of people believe that if they cannot understand everything, little bits and pieces in the model and how it works, they shouldn't use it. That's the other problem. There are lots of examples out there, but I'm going to use airplane example. We're all using airplanes to do travels very, very frequently around the world, but actually we don't know much about how it is being built. We cannot build it ourselves. Definitely, there are lots of complex little bits and pieces that we will never understand. Yet still, we get into the plane, happily drinking the coffee and waiting to arrive. That's the thing. Why it's happening? It's because we trust airplanes to be safe just because it has been tested a lot. It has been tested a lot frequently by people that are trusted as technical people, right?
Steve Nouri: 00:47:03
That's the same thing for AI. We need to trust the people that there are technical, that they test it, and then we need to use it as a tool. I have seen, unfortunately, a lot of people that they don't trust it. They don't trust it based on the example and based on the test. They just say, "I need to understand exactly the bits and pieces until I use it." That happens in many industries. For example, I used to work in a FinTech company and financial advisors, specifically, they were asking how this AI is predicting blah? How is AI predicting how much you should invest? I need to see a formula that has ABCD in it and then see how it came up with the formula. I need every detail and I'm like, "It's not going to happened."
Steve Nouri: 00:48:00
You can not have such a thing in that detail. You can see the result and see if that works. If that doesn't work, then the data is wrong and you can just go into some other phases to make it right. But if you just want to shut it down, just because you don't have all these details, then you're missing out. But that was the major thing that I mentioned, because I was working in that industry and I was dealing with many people across different industries, that they were not happy if they cannot understand the details in the model, or how these predictions or the result has been generated. I think we will get better as we use AI on a daily basis, and we understand it was AI that we were using all the time. So, that's good. That's the education part that I talked about.
Steve Nouri: 00:49:07
As we can see that people are using Netflix, and Netflix recommends movies that are particularly interesting to you, and that's AI. And if people understand that's AI and they're using it daily, then they feel a little bit more comfortable, that oh, okay, it seems it's working. And Alexa is talking to me and giving me some good advices. Let's say they're using some fancy Tesla models, and they're using these autonomous functions and features, and they're enjoying that and they're trusting it, so they're getting there. But I think it was last year that I talked about it. I wasn't so sure that people are yet there, ready to accept it in the workplace as a replacement.
Kirill Eremenko: 00:50:13
I see. I see. So, if my understanding is correct, then it's kind of like taking your airplane analogy. If we look back... Now, everybody's flying in airplanes and enjoying it. Well, barring coronavirus, of course. But when airplanes were first created by the Wright brothers a long time ago, people were probably very skeptical of getting on the first plane, and I remember the elevators. When the first elevators were created, people were like, "Whoa, this is not going to... This is too dangerous. I don't want to get in this thing," and people were skeptical, but then as time passes, people will become more and more comfortable with it. Is that about right, as a summary?
Steve Nouri: 00:51:03
Exactly. Yeah, that's right. That's why I'm always emphasizing on education, and that's why I'm sharing all of these examples daily, because I think that's the only way to get out of this. That's the only way. If people understand how AI is impacting their life and how it is used, then that's the moment that they will start trusting it.
Kirill Eremenko: 00:51:35
Gotcha. And this is actually a good segue to our final part that I want to talk about on this podcast, which is personal brand, like creating a personal brand. And I know this is something you're passionate about and you've done extremely successfully in terms of creating that brand and maintaining it, that you do help a lot of people with your posts, and I think you're a Forbes contributor as well. So, like your recent article was about where can you find great NLP courses for 2021, which is another form of aggregating knowledge and sharing with others. Why would you say it's important for everybody to look into building a personal brand in this space?
Steve Nouri: 00:52:25
So, first of all, everybody in the digital world needs to have an identity and needs to connect and interact with people. We are social people, and in a normal world, you talk face-to-face and people will know you based on the interaction they have with you daily, but in the digital world, it is not possible for you to talk face-to-face with everyone. I cannot literally have a long conversation with a couple of hundred thousand people. So, what I can do, I can... In order to understand the others, and in order to... them to know about me, I can make a personal brand. So, by personal brand, I mean the perception of you and how you would like people to understand you as a person.
Steve Nouri: 00:53:32
And this is what I'm really passionate about, because I would like people to know me as a person who inspires and empowers people in AI and data science. That's the main thing I would like to communicate, and I'm actively working in both sides, so I share a lot of interesting projects, innovations for inspiration, but then I share things like the Forbes article that will tell people how to learn, where to learn, how to up-skill, and that's where empowering. So, that's kind of the way I'm focusing on my personal brand and the way that I am showing how I'm contributing to the society.
Steve Nouri: 00:54:25
And it is especially important for beginners because they want to find a job, and if you're looking for a job, these days, hiring managers highly likely will search about you and they will find your digital footprint to understand about you. And that's where your personal brand will show. Because when you have a lot of GitHub contribution or you have a lot of Kaggle competition that you've participated... Doesn't necessarily need to be winning a Kaggle competition.
Steve Nouri: 00:55:14
You can just participate in a lot of them, and that's also showcasing your technical knowledge, and also, it will be as kind of a softer skill, showcasing your softer skills, that you are a person that will push, that would like to learn, would like to engage, and these are all the things that will come up out of your digital footprints. And by putting them together, you're making a brand for yourself, that how you would like those HR manager to see you, as a data science enthusiast, for example. If you have written a couple of blogs, then that's a plus. These are all little bits and pieces. It's what I'm doing right now on social media. I'm encouraging people to engage more, to be more active, helping each other, and also to make contributions to open-source because these are the way that you can showcase your knowledge, your understandings, and definitely that will hugely help you to find a job.
Kirill Eremenko: 00:56:38
I completely agree with that. And here's just an example, we were hiring a data analyst at SuperDataScience a few months ago, and so for all of the candidates, first thing, I would not even look at their CV. I would go to their LinkedIn, see what they're up to, what are they doing? And the candidate we ended up hiring, one of the biggest things that stood out to me, I clearly remember to this day, is that she ran workshops. She was very passionate about the teaching the stuff she knows like data visualization. She wanted to give back. And she was not only doing the work, but she was teaching others. And to me, that's a huge indication that you know what you're talking about if you're able to teach it. Because if you don't really know, if you're shaky on your grounds, then you're probably not going to be confident enough to go out there and teach it. It's going to be like a taboo for you. You first need to learn really well. So for me, that was like, boom, a huge plus, as you say. Right away.
Steve Nouri: 00:57:39
100%. Yeah, that's something that people are understanding more and more these days that they need to be more active on social media. They need to be more active on contributing to other projects and helping others as well. And that's actually great. That makes the community to work together and learn from each other. And I'm actually doing very soon I'm going to make some courses about it. Run round some workshops about personal branding and how to be more active on social media, how to make your social media profile pop, and also engage more with your audience, have the focus. And these are the stuff that I think it will be interesting for everyone, not only job seekers. It can be interesting even to professionals, especially to people who would like to accelerate in their career.
Kirill Eremenko: 00:58:44
Absolutely. Absolutely. What would you say... So, first of all, what's the best way for people to get involved in these workshops or courses when they come on?
Steve Nouri: 00:58:56
So around November, I'm going to announce the workshops and courses. So just stay tuned and search my name on LinkedIn. There are not many Steve Nouris LinkedIn that's easy to find. Follow me on LinkedIn. I will share it very soon. Apart from that, if you follow, you will get a lot of other interesting stuff, hopefully. So, stay tuned for courses and workshops. Workshops, so some of them are going to be free as well, and hopefully people can get something out of it and learn something that will help them to accelerate.
Kirill Eremenko: 00:59:43
Awesome. Are you going to run any of those virtually?
Steve Nouri: 00:59:46
Yeah, so all of them going to be virtual because of just having my followers are across the world and my connections are scattered around the world. It's very difficult to run it in person. I would like to scale it up and help as many as I can.
Kirill Eremenko: 01:00:07
Okay. Fantastic, fantastic. Well, that's a really inspiring thing for people to be part of. I don't think I've seen many workshops on how to build a personal brand. I think a lot of people are actually understand the value in it and want to do it, but don't know where either to get started or maybe feel like they're not expert enough. Like, "Who am I to share things or contribute?" Like, "I need to learn data science very well first, and then contribute." What would you say to people who have that mindset right now?
Steve Nouri: 01:00:47
Yeah, that was actually something that I was struggling with when I started. I was always thinking the same. I need to be an expert in some areas before start talking about it. And then I learned that I can just share what I know. It doesn't need to be perfect. There's no such a thing as a perfect. If I want to wait to be perfect, I will die without being perfect because there is no such a thing. So just don't think about that. Just think about how you can add value if it is just... This is the last book I read, and I liked it. That can be a way to help the community, that you can write two sentences about that book that you liked about it, and that might help someone to find the book that they were looking for so long. And it just builds up from there.
Steve Nouri: 01:01:46
I started sharing my passion about data science because when I was lecture at university, I was getting a lot of questions, repetitive questions, from students about data science. For example, "What is the next step for me? What is the best way to learn Python? Is there any free book about machine learning and [inaudible 01:02:15]?" So it was all coming up again and again. I was thinking to myself that maybe that's a question that more than five or six students have in the whole world, maybe that's something that I can help others as well. So I started posting about it. I was like, "Okay guys, this is a free," I don't know, "book that I found about machine learning. It's from five years ago, but is still relevant. Let's see how it goes." And then boom, people liked it.
Steve Nouri: 01:02:47
They were started following for the same content, and I started engaging with them, and giving them more of the same material. So that was the way that I got more involved and be proactive. And then I learned how to add more value by adding some interesting projects because only learning is not going to be enough. For my audience and my connections, they need to also understand how to use those learnings and apply it in something. So then I started sharing the applications and interesting projects, which will make people excited to see what would happen that you learn computer vision. You can identify dogs and cats, for example, and that's something great for someone that has never done any programming. Just 10 lines of code and boom, it's there.
Kirill Eremenko: 01:03:49
Fantastic, fantastic. That is very inspiring, and I think it just takes one step. The first step, just share something and you'll see that it's not that scary, and if you're going to help one person, that's amazing already. Steve, we are running out of time. I want to thank you for your contribution to everybody listening to this show. It's been a huge pleasure. And before you go, where are the best places to find you? You already mentioned LinkedIn, I will definitely include that in the show notes. Is there anywhere else our listeners can find and connect with you, maybe follow you and find out about the projects that are coming up?
Steve Nouri: 01:04:34
Yes. HackMakers is another interesting project of mine. I'm involved as an advisor and they're running some workshops and events very soon. That's another place to find me, but I'm only active on LinkedIn, just because I cannot afford to spend any more time on social media, I'm already spending two to three hours a day.
Kirill Eremenko: 01:05:05
Oh, wow.
Steve Nouri: 01:05:08
Because I receive more than 50 messages a day and I need to spend some time to get back to them, and also curating the content and finding interesting resources will also take time. So I'm only focusing 100% on LinkedIn to add value to my audience. And I keep it to my focus area, which is AI and data science, and if you like it, I would like to get connected to you and talk to you about the same passion.
Kirill Eremenko: 01:05:42
Awesome. Awesome. That's very cool. Love the focus. And one final question for you, what's a book that you can recommend to our listeners?
Steve Nouri: 01:05:50
I've recently received a book from a friend of mine, Andriy Burkov, if you guys follow him on LinkedIn. He shares a lot of interesting photos and also interesting facts about data science and AI. The book is called Machine Learning Engineering, and it was published recently. That's interesting. I'm still reading it.
Kirill Eremenko: 01:06:18
Awesome. Machine Learning Engineering, by Andriy Burkov, another great person to follow. I follow him and some photos he shares are cool and some memes, he shares memes sometimes as well.
Steve Nouri: 01:06:29
Yes. Exactly.
Kirill Eremenko: 01:06:29
Awesome. Well, Steve, thank you so much for coming on the show, it has been a huge pleasure and I'm very excited about all the things that you're doing for the world of AI and data science through ACS and personally as well. It's a pleasure to know about it and inspiring. I'm sure it's been inspiring to everybody.
Steve Nouri: 01:06:52
Thank you very much, Kirill, for having me.
Kirill Eremenko: 01:06:59
There you have it, everybody. Hope you enjoyed this podcast and this conversation with Steve, and I definitely did. I definitely got some valuable takeaways and I loved the flow of our conversation and how we covered off quite a few different topics in the space. My personal favorite part was Steve's experience and stories of hackathons, how he's participating in many of them and how he's run several already with the Australian Computer Society. It's just exciting and inspiring to see how many people are participating, in the thousands, so they ran the largest one in Australia. Then the global one, they had also thousands of people respon, on the January hackathon, they're aiming for 10,000. So make sure to stay in touch with Steve, follow him and see when that hackathon comes around. I think it's a great opportunity to add something to a portfolio, to practice, to meet people, and why not? Why not participate in a co-hackathon for data sciences like that?
Kirill Eremenko: 01:08:05
So as usual, you can find the show notes for this episode at superdatascience.com/409, that's superdatascience.com/409. And there you can find the transcript for this episode and materials we mentioned on the show, plus the URLs for Steve's projects and his LinkedIn, where you can stay in touch with him and see what things he shares. He shares some pretty cool things on LinkedIn.
Kirill Eremenko: 01:08:34
And if you enjoyed this episode and you know somebody who's starting out into the space of data science and artificial intelligence, then share the love. Spread this episode with others. It's very easy to share. Just send them a link, superdatascience.com/409. And on that note, thank you so much for being here today, really appreciate you spending this hour with us. And I look forward to seeing you back here next time, until then, happy analyzing.
Show all
arrow_downward