Can artificial intelligence grow safely?

This is an audio transcript of the FT News Briefing podcast episode: ‘Can artificial intelligence grow safely?’

Marc Filippino
Good morning from the Financial Times. Today is Friday, October 6th, and this is your FT News Briefing.

[MUSIC PLAYING]

We might be in for a slowdown in today’s US jobs report. And the bond market went haywire this week. Plus, popular AI bots like ChatGPT are starting to look and act more like humans. But are we ready for that? 

Madhumita Murgia
We still don’t know why and how they work. We don’t know why they produce the responses they do, and we actually find it quite difficult to control them. 

Marc Filippino
I’m Marc Filippino, and here’s the news you need to start your day.

[MUSIC PLAYING]

Now, let me be clear before I say this next thing. These are just estimates. But US jobs growth is expected to have slowed in September. Bloomberg surveyed a bunch of economists. They projected that today’s data release would show US employers added 170,000 new jobs in September. That’s a step down from August’s increase of 187,000. The US Bureau of Labor Statistics will release the actual figures at 8:30am New York time.

[MUSIC PLAYING]

Central banks say that interest rates will have to remain higher for longer to get inflation under control. But the idea of a lengthy period of high interest rates is making investors nervous. Stock markets have tumbled and long-term bond yields have soared over the past few weeks. I’m joined by the FT’s markets editor, Katie Martin, to make sense of all this. Hey, Katie. 

Katie Martin
Hey. Hey, Mark. How you doing? 

Marc Filippino
I’m doing all right. I have this pink newspaper thing in front of me. I don’t know if you can hear it, but on the front page it says: Bond sell-off deepens as high rates prospects spook markets. 

Katie Martin
Yeah. 

Marc Filippino
Why, like, why is this happening even though central banks are levelling off on interest rates and inflation seems to be levelling off? 

Katie Martin
Look, it’s a good question, and it shouldn’t be so difficult to answer, but it is in a way because, you know, first of all, central banks have been really clear for months and months and months that they’re not joking about this whole inflation malarkey. They’re going to push rates up until they vanquish it. 

Marc Filippino
And you and I have talked about why investors haven’t been listening. There’s seemed to be this disconnect for a while now. 

Katie Martin
Yeah, just la la la, don’t worry. The Fed will go back to its old ways and cut rates as soon as we hit a spot of trouble. And the Fed has been pretty clear that that’s not going to happen. But nonetheless, the market just keeps testing the Fed, I guess. What’s actually happened is instead, first of all, the market has been really slow to really get this higher for longer message. So the Fed’s been saying higher for longer, and the market has been wronger for longer and just tried to take the other side of that. The other is that what we’ve seen since the summer is a massive kick higher in oil prices. And so suddenly people have got worried that inflation is not going to sort of gently sail lower like we’ve been hoping. You know, maybe there could be a resurgence here. And so what we’ve seen is another big step down in bond prices, a big step up in yields. You know, we’ve had yields on 10-year US Treasuries and 30-year US Treasuries hit the highest levels since 2007. We’ve seen some really big moves. Thankfully, as I speak at least, this has calmed down, but it has been one of those weeks and one of those months where people have thought, hang on, markets are really not doing what I thought they were gonna do, and this is quite unnerving. 

Marc Filippino
So like you said, we’re starting to see some relief. Katie, is this the light at the end of the tunnel? I mean, what does this all mean? 

Katie Martin
So what I think we all hope it means, because we don’t like excess volatility because it’s dangerous in markets, what we all hope it means is that this is becoming self-correcting. And also the idea is that if interest rates get too high, then that’s too damaging for the real economy. So there’s less demand for energy. So that pulls the oil price down. But the big picture remains that higher for longer is finally sinking in. And all different types of investors — whether they’re retail investors, whether they’re big institutional investors — are figuring out how they’re going to deal with that because this is going to be with us, you know, not just for the next few months, but for the next few years. And so you’ve got to get ready. 

Marc Filippino
Katie Martin is the FT’s markets editor. Katie and our colleague Ethan Wu had a great chat about bond yields on the FT podcast Unhedged earlier this week. We’ll throw that link in the show notes. Thanks, Katie. 

Katie Martin
No problem. 

[MUSIC PLAYING]

Marc Filippino
So are AI chatbots getting smarter? Last week, OpenAI, the company behind ChatGPT, announced that it was rolling out a bunch of new features. They promised users a new and more intuitive interface. Here to walk us through some of the updates and what it all means for the future of AI is the FT’s artificial intelligence editor Madhumita Murgia. Hey, Madhu. 

Madhumita Murgia
Hi there. 

Marc Filippino
So walk me through what’s new with ChatGPT. What can the bot do now? 

Madhumita Murgia
So the latest releases are pretty interesting because previously the way that people have been interacting with ChatGPT is by using text or writing. But what they’ve released recently is making it multi-modal, which means you don’t only have to ask questions using words or text, you can ask them using pictures, and ChatGPT can respond using images as well. So you could upload a picture of a cake and say, can you give me a recipe to recreate this image, for example. And it could analyse the image, figure out it was a chocolate cake and find your recipe for it. The other thing they announced also is that you can now talk to ChatGPT using voice, and it can also respond verbally. So you can really have . . . converse just purely in voice without typing anything, which I think is also a big change in how we sort of interact with AI systems. 

Marc Filippino
So, Madhu, it kind of sounds like Siri plus Alexa plus a chatbot all rolled into one. 

Madhumita Murgia
Exactly. You know, the way they sort of pitched it was that it can now see and hear and talk to you. So it’s just a much more natural . . . like the way we would interact with one another. 

Marc Filippino
How have other tech companies responded to these new updates? Are they doing their own updates? 

Madhumita Murgia
Yeah, I mean, we’ve already seen over the last year really an explosion in the number of chatbots on the market. But even just in the last couple of weeks, Meta, which is the owner of Facebook and Instagram and WhatsApp, has really jumped into the fray. But what they really announced is their own AI assistant, which will be something that you can access through all of the different Facebook platforms and ask it to help you out. But also this whole bunch of AI characters which you can talk to, so that’s just much more of like a fun entertainment thing. So you can have all these different characters, like Snoop Dogg and Kendall Jenner and a whole host of celebrities who sort of voice different characters that you can chat with and play with and sort of have fun with. 

Marc Filippino
OK, this sounds fun at all. And I’m not gonna lie. I think I would be entertained talking to a fake Snoop Dogg. But are there still concerns with AI and are they being addressed? 

Madhumita Murgia
Yeah. So I think they just haven’t solved the problems that come with this. As you can imagine, you know, if you just have an agent, a computer agent that you can talk to that will tell you anything, there are a whole bunch of risks and dangers that come from that. They might say harmful things, dangerous things, racist, sexist things. They might give you false information. So it’s really dangerous because there’s no sort of filter between us and these programs. And, you know, I’ve been speaking recently to some of the CEOs of these companies, like Anthropic’s CEO who came out of OpenAI, and they say that we really have no idea what’s happening inside these models. We still don’t know why and how they work. We don’t know why they produce the responses they do, and we actually find it quite difficult to control them. So I think the safety aspect is way behind the actual technical development and the adoption that we’re seeing of AI in the market. 

Marc Filippino
Madhumita Murgia is the FT’s artificial intelligence editor. Thanks, Madhu. 

Madhumita Murgia
Thank you. 

[MUSIC PLAYING]

Marc Filippino
You can read more on all of these stories at FT.com for free when you click the links in our show notes. This has been your daily FT News Briefing. Make sure you check back next week for the latest business news. The FT News Briefing is produced by Kasia Broussalian, Sonja Hutson, Fiona Symon and me, Marc Filippino. We want to say a special thank you to Monique Mulima, our intern. It’s her last day here at the FT. Thanks so much for all your hard work and good luck on your next adventure. Our engineer is Monica Lopez. We had help this week from Persis Love, David da Silva, Michael Lello, Peter Barber and Gavin Kallmann. Our executive producer is Topher Forhecz. Cheryl Brumley is the FT’s global head of audio, and our theme song is by Metaphor Music. 

[MUSIC PLAYING]