OpenAI + Journalism = Never Too Late To Be a Jedi
In the news, NYTimes suing OpenAI .?
The NYTimes is upset. You can relate. No money paid. No attribution. Not nice.?
If you believe the past repeats itself as Karl Marx suggested, then we’ve actually seen a very similar scenario before. Meta (then Facebook), almost a decade ago in 2015 launched Instant Articles (IA). It was an initiative to host publishers' content on Facebook and keep users within the app to read it. Despite using valuable news articles and hard work from journalists, Facebook did not pay a lot (or at time nothing at all) to publishers, and stopped sending any traffic to publishers. Facebook benefitted as their app rendered a lot faster, and on paper the user experience was better. In the end, it broke, and funny enough (or not), NYTimes was among the first publishers to get out of Facebook Instant Articles. Many publishers followed.?
10 years later. Where does that leave us with OpenAI dynamics with publishers, the open web and supporting journalism??
OpenAI has their own argument for why they are right to use publicly accessible news stories to train their AI, and maybe they are right. Maybe crawling the web the way they do is no different than the way Google crawls the web to enable search.
However, like many “debates,” the majority of times it doesn't really matter who is right, but rather what is your desired outcome, what is the culture you're building, your identity, your values. How does OpenAI want to be remembered? they could evolve to be a force of good, a ‘Jedi’ to support journalism, support high quality content, and the open web. Or they can decide not to do it, and evolve to be a Sith lord, not pay and not attribute.?
If the top 100 sites in the world (Wikipedia, NYTimes, Reddit, etc) block OpenAI or even demand it deprecates their data from their past crawl, then OpenAI would be much less valuable.?
领英推荐
On the flip side, GenAI is such a revolution, perhaps one of the biggest things we’ve seen since the internet started - can it really be ignored??
I bet that OpenAI will do the right thing. The Information reported OpenAI is considering paying up to $5M to license content from publishers to train its AI. And from my point of view, OpenAI should pay whatever publishers want and a lot more. As opposed to Facebook which is a 100% advertising company, OpenAI has the opportunity to put their technology at the hands of hundreds of millions of users through partnerships with enterprise accounts all over the world, and charge for it. In fact they can charge for it a lot, if you compare how much enterprise accounts pay for cloud services, this can easily be a $100B a year revenue to OpenAI. So if that’s the case, does it matter if publishers are being paid a lot? Not to mention, OpenAI can be remembered to be on the right side of history, as opposed to what Facebook did to publishers and the open web (not nice).
Given the relationship between Microsoft and OpenAI, I think Microsoft is a better life partner to journalism than Facebook ever was, or in fact wanted to be. That makes me think that OpenAI will do the right thing, and where Facebook failed, Microsoft will succeed in making the open web and journalism stronger, paying editorial teams for the important content they create, and the payment they deserve to get.
I also remind myself as a parent of three how important this question is. We (humanity) need a very very strong, thriving open web and journalism. With social networks endangering the future of our children, surfacing so much hate, and fake news, god help us? - we need editorial teams like we need oxygen. I wrote about it on CNBC here https://www.cnbc.com/2023/07/10/op-ed-taboola-ceo-says-hed-rather-let-kids-play-with-ai-than-tiktok.html ?
In summary - I’m optimistic.?
I bet OpenAI will do the right thing. It is never too late to be a Jedi.?
What’s your “Once Upon a Time”? Working with companies, brands, and brand leaders - world leaders, as well - to construct narratives that harmonize with this moment and anticipate the next.
9 个月Hi Adam. Because I am a pedant, I will correct you and say that Marx commented that history repeats as farce. And perhaps it can be argued that there is a farcical component to Open AI, because AI uses the corpus, draws upon the combined body of whatever passes as human knowledge, which is part of the long comic tradition of farce as playing with reality. The process of training AI on existing content has no direct precedent, at least as I can see it, in the history of technology's relationship with existing IP. Perhaps if Guttenberg had been sued for taking illuminated monastic manuscripts and turning them into printed output, that might have been an existing referent. My view, as a writer and creative being, is that there is nothing more precious than what the human soul, in kinetic conversation with the world, is capable of creating. We don't create to enable Sam Altman to raid, cannibalize, harvest and appropriate content to train his AI engine and then monetize that in a repulsively unrestrained captialist orgy. I get that 23andMe can "train" its engine on anonymized DNA, but I find that equally reprehensible. Sent from my iPad
Client Solutions Director, ANZ | Pulsar
10 个月Thanks, Adam Singolda! May the force be with good-quality journalism. Enjoyed reading the historical parallels with Facebook's Instant Articles. Can't decide if I'm on Team Optimistic yet, but indeed, it's fascinating to observe the ever-changing relationship between technology, AI, and journalism.