AI Business Podcast

AT&T's chief data officer Andy Markus joins AI Business Editor Deborah Yao to discuss how the telecom giant uses AI to manage petabytes of data across its global network and deliver key insights for business managers. He also talks about the rise of AT&T’s citizen data scientists as enabling the democratization of AI across the company, which can boost business value by 5x or more – and reveals whether he thinks AI can ever be sentient.

In this episode, Ondrej Burkacky, global co-lead of McKinsey's semiconductors practice, joins AI Business Assistant Editor Ben Wodecki to discuss the ongoing chip shortage. Two years on is the end in sight? Is the crisis over? Will we ever be able to purchase a PlayStation 5?

In this episode, Peter Stone, executive director of Sony AI, joins AI Business Assistant Editor Ben Wodecki to discuss this year's RoboCup and the AI that can beat you at Gran Turismo. 

Dell UK CTO Elliott Young talks to AI Business Editor Deborah Yao about keeping AI models accurate by avoiding AI drift, what measures to put in place for proactive prevention and why avoiding it is key to AI performance. Also, he reveals what keeps him up at night.

The AI Business Podcast returns - with a new lineup! AIB stalwart Ben Wodecki gets a new title and some new friends - visionary business and tech journalists Deborah Yao and Mukul Pandya join the pod's roster to bring you some insightful and engaging conversations about AI, machine learning, data, business intelligence and much, more more — new episodes coming every other week!

This week, the editorial team shares the most profound things we have learned about AI over the past 37 episodes. The reason is simple: Tien is departing for pastures new, Max is making a lateral move to Data Center Knowledge, and Sebastian…  Is doing Sebastian things. Which means this will be the last time the podcast will feature the current cast. Ben is not going anywhere, by the way; the podcast is in his hands, and you can look forward to many more episodes following a short break! In this episode, we try to share the most important things we have learned: Max will scream that AI is just software, and software is still eating the world; Sebastian will bemoan the fact that despite all the potential for innovation, the entire AI market ended up being dominated by the same three or four familiar names that already control most of the world’s IT infrastructure. Ben will highlight the importance of training, and the fact that for all the assumed intelligence, AI systems can only do the things they are painstakingly taught to do – but he’s hopeful about some of the applications. And Tien will wonder about the sheer variety of things people call AI – how do you make sense of something that applies in both pharmacological research and sewer management? We also cover: Body odour! Kwisatz haderach! Reporting on weird stuff! Potential future collaboration! For the last time, you can find the people responsible for the podcast online: Max Smolaks (@maxsmolax) Sebastian Moss (@SebMoss) Tien Fu (@tienchifu) Ben Wodecki (@benwodecki)

This week, we couldn’t resist talking about Meta Platforms Inc. – and all the dumb things Facebook has done before. We also look into carbon-neutral AI: Northern Europe might be a cold, dark and unforgiving place to live, but it’s a perfect location for hot and power-hungry data centers that run AI training workloads. We start with Meta – a new name for a holding company that owns Facebook, along with Instagram, WhatsApp, and a bunch of other tech subsidiaries. With its flagship platform facing a string of PR disasters, the people in charge have decided that it was the right time to deliver on the outlandish ideas promoted by The Zuck over the past six months. Meta’s big bet is virtual reality – something that countless other businesses have tried, but invariably failed, to take mainstream. Is it any wonder we have decided to drink this week? It is true, following its acquisition of Oculus, and with video game legend John Carmack on board, Facebook Meta has some of the best chances of actually delivering on the promise of VR. At the same time, the company’s 3D wizardry clearly needs more work – who renders fire like this in 2021? There is nothing impressive or ‘life-like’ about the tech demonstrated by The Zuck. Next, with the COP26 summit in Glasgow in full swing, we look at how AI builders can make their models more sustainable. We discuss the specific example of Neu.ro, a company that created a zero-emissions AI cloud in partnership with atNorth (erroneously referred in the podcast as NordDC), operating from Iceland, where servers are powered with geothermal energy. atNorth is not the only company doing this: in one of our previous podcasts, we looked at the example of Verne Global, another data center operator to feed AI models with renewables. Sweden, Norway and Finland are all making similar data center pitches – and it makes good commercial sense, with the Nordics enjoying some of the cheapest electricity in the world. We also cover: Second Life! PlayStation Home! The eventual death of the planet! Can you find all the f-bombs in this episode? As always, you can find the people responsible for the podcast online: Max Smolaks (@maxsmolax) Sebastian Moss (@SebMoss) Tien Fu (@tienchifu) Ben Wodecki (@benwodecki)

We delve into the irrational world of WorldCoin – the unlikely enterprise backed by child prodigy Sam Altman that wants your eyeballs. To scan them. Using a shiny metal orb. In exchange for some virtual currency. This is not an elaborate joke – investors have reportedly valued WorldCoin at $1 billion, and the eyeball harvesting operation is in full swing. There are 30 prototype orbs in the field, according to TechCrunch, scanning up to 700 eyeballs per week. They want thousands of orbs, and eventually – a billion eyeballs, each representing a singular and unique cryptocurrency user. But before we get to the orb, we discuss the US military, and its plans for a robotic tank unit. For now, the US Army Futures Command’s Next Generation Combat Cross-Functional Team is putting together a bunch old M113 armored personnel carriers, and is making them shoot each other. The likes of BAE Systems, General Dynamics, and Oshkosh Defense have all been tasked with designing prototype unmanned combat vehicles for the US Army – the only question is how quickly these will appear on the battlefield. As always, you can find the people responsible for the podcast online: Max Smolaks (@maxsmolax) Sebastian Moss (@SebMoss) Tien Fu (@tienchifu) Ben Wodecki (@benwodecki)

This week, the editorial team at AI Business is looking at Facebook, suffering the fallout from both its massive service outage and the testimony of Frances Haugen before the US Congress, which gave us a glimpse of the dark science taking place inside the company’s software development labs. And it sure ain’t pretty. There’s a stunning lack of morals at the core of Facebook – these guys can be linked to destructive social trends just as surely as tobacco companies can be linked to lung cancer. There’s almost no chance of meaningful regulation – but forcing senators to say ‘algorithm’ is a step in the right direction. Haugen is testifying before the UK Parliament later this month. We also delve into the tyranny of feeds – which can’t seem to adapt to our preferences. The promise of personalization that accompanied countless data collection efforts was a lie. Next, we delve into the world of pharma: where virtual dogs will hopefully replace live animals for drug testing. Meanwhile, AI for drug design is raising tons of cash, with British startup Exscientia netting $305m in its IPO in New york, along with $160m in private placements. This follows a $400m funding round for XtalPi, a similar company primarily based in China. We also cover: Lawyers! Newcastle United! DeepMind! Smash that like button if you like Squid Game. As always, you can find the people responsible for the circus podcast online: Max Smolaks (@maxsmolax) Sebastian Moss (@SebMoss) Tien Fu (@tienchifu) Ben Wodecki (@benwodecki)

This week, the editorial team at AI Business looks at the latest developments in video games – which are making more money than movies and music combined, but remain something of a second-class citizen of the entertainment industry. Machine learning is coming to video games, and while this brings plenty of positive developments, we are going to discuss something unpleasant: the treat it poses to competitive online multiplayer. With computer vision tools now easily accessible, a new generation of cheating techniques is ruining games while remaining virtually undetectable by common anti-cheat software, because computer vision doesn’t need to need mess with game files – all you need is the video output. This is especially damaging for popular games like Call of Duty: Warzone or Fortnite, where aiming skills and fast reflexes often mean the difference between victory and defeat. Look out for Ben’s feature on the topic later this week! We also cover: Battlefield 2042! Unintended consequences of progress! Teenage angst! As always, you can find the people responsible for the circus podcast online: Max Smolaks (@maxsmolax) Sebastian Moss (@SebMoss) Tien Fu (@tienchifu) Ben Wodecki (@benwodecki)