Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: GPT-3 and the Future of Advertising
14 points by georgehill on Dec 25, 2022 | hide | past | favorite | 20 comments
Hello HN,

I am just curious. There are a couple of products (You, Perplexity.ai, and Neeva (soon)) that have recently launched a feature that summarizes articles and spits out answers. I'm wondering how creators, particularly news outlets, will get paid if search engines start taking their traffic?

Maybe completely new business models will need to be implemented?




My brother writes sports copy for a local news outlet.

I'll just paste the text he sent me when I asked basically the same question.

"I'll worry about this when AI can watch baseball and make sense of a manager getting ejected."

I think "news summaries" are going to get blown up by AI reading all of the other news articles that exist about something, but ... someone has to actually write the first draft. Back to sports: we're a long way from an AI watching a ballgame and understanding why, exactly, a manager had his pitcher hit the next batter, or why a certain homerun was so crucial.


With live data feeds for sport events already available for years, I think this is either in development or maybe already in use.


will AI really need to watch the game or could it just listen to a television or radio broadcast?


You could literally do this today if you have a good source for broadcast streams, connect it to Whisper or something to create a transcript, and use a version of say text-davinci-003 trained on such transcripts and/or subsequent news reports with a prompt asking for a short news article based on the broadcast and post game news conference transcripts.


True, AI can take any human communication and repurpose it, if it has access.

People keep forgetting the shortcuts AI can take. ChatGPT takes shortcuts to another level. Bypass understanding the meaning and go straight to the output.


Gpt3 is going to kill Google.

News outlets have already died. What exists today is not news. Those aren't journalists. The AI can already write better.


Gpt3 is definitely not going to kill Google.

Gpt4 might though, but probably not, large companies are hard to kill and Google is at the forefront of this kind of research so likely they would counter with their own language model. Imagine if the thousands of data scientists and billions of compute that currently goes towards improving search gets retasked with making language models, that would happen really quickly if Googles main source of revenue gets challenged.


They won't because that's just how transitions happen. They would need to completely change how their interface works and pivot from their current customer base.

Someone will figure out a way to make a hyper localized chat bot which sends referral links.

I'm not too sure of the different levels of machine learning but and got but gpt3 chat and services like it are time saving.

I was able to watch a non technical person buy a VM from linode, deploy debian, setup WordPress behind nginx and setup ssl offloading.

Google would never be able to help. The person would just quit from overload.


Google has already pivoted in that direction, they already try to answer questions directly in the response window instead of just providing links. All they would need to do is make a language model provide those answers instead when it is sure enough.

And that would also solve other issues such as trying to validate those answers. The language model could get the first response page and base its answer on those, basically summarising the first 10 pages making it way more reliable than ChatGPT is now.


They won't do it though. They can't risk the cash cow they have.

Think Digg when they came out with the current Reddit design.


It probably also means that the AI can write better advertisements than what currently exists.


anybody remember Summly[0] from a decade ago?

> Yahoo has announced that it will be acquiring Summly, a startup that aims at summarizing web content into a digestible format more fit for mobile devices. In a nutshell, what Summly does is let you pick your news sources from a set of pre-packaged categories or your favorite website. From there the app will show you the latest stories summarized in up to 400 characters and presented in a cleanly designed interface.

I don't think this did any damage to the news publishing industry. Article summaries have existed forever: most of the free blogosphere does anything besides add their 2 cents to stories broken by the larger news outlets.

[0]https://www.techspot.com/news/52054-yahoo-acquires-news-summ...


The future of advertising is this. GPT-3 will be trained on all content you have ever produced, will build a model of you, and generate an ad copy (and media) that will move you personally at this particular amount of time to buy the service.


Advertising is a doomed business model. It's not something the end-user wants, yet the model relies on the end-user consuming it. Obviously countermeasures will emerge, such as ad-blockers or running the content through something like GPT-3 to strip out the spam and only leave the actual content the user wants.

The only way out (especially if advertising is no longer paying the bills) is to align the incentives and get end-users to pay for the content directly. If ads no longer pay the bills, there will be an incentive for all publishers to work together on a frictionless, interoperable micropayments service so end-customers can pay to replace the lost ad revenue. Paywalls will be erected to prevent search engines from accessing (and then republishing summarised versions) of the content for free.

Some parasitic trash will definitely go out of business but that's for the better. There's currently a lot of content and companies out there who only survive because people can't ask for a refund on an ad view - by the time the user realises they've been tricked into clicking on trash, the creator already got paid. This model wouldn't work in a micropayments-based world.


Unless you’re running the models yourself with your own queries, you’ll just end up consuming model output that has been prompted to inject advertisements.


Advertising will be there for ad long as the need to sell things is there.


Advertising will remain in places where it can’t effectively be blocked. The web is increasingly not that. Print & display advertising, sure.


Blocking is opt-in, becoming increasingly more difficult, and next to impossible in native apps (which is why every ad-supported site like Reddit constantly spams you with “it looks better in the app”, and deliberately spoils the web experience).


At this time, most agencies are largely ignoring the advancement of generative AI technology. This is likely due to most of the time in advertising being spent on developing campaign concepts and generating ideas. AI tools like GPT-3 and ChatGPT still struggle with coming up with high-quality campaign ideas. Our agency is using GPT-3 for summarizing and condensing briefing reports, writing "Schweinebauch" (which are equivalent to press releases, technical copy for automotive ads, and display ads on the web), and ChatGPT for creating target group personas that our creatives can engage with. For Art Direction, we're employing self-trained, stable diffusion models for storyboarding, style frames, and photo bashing layouts, which has drastically increased our prototyping speed. In the past, it was a running joke that if an Art Director said they needed thirty minutes, they would actually need two hours. But that’s no longer the case as this approach makes pre-production so much faster. I’m sure it has the potential to bring about real change to the industry, but for now, AI is just a useful tool that can’t fully replace internal resources. We’re seeing a drop of about 70 percent in the number of layout illustrators and stock images we need to book though. The main consequence of this, however, is that our creative directors are now having to deal with a much higher workload because they’re being presented with substantially more content to evaluate. Last year, we learned that copywriters and art directors who had a strong foundational understanding of their craft and a good grasp of their sector’s professional jargon produced much more impressive results with the help of AI than the ones who primarily relied on tutorials and trends for their work.


> We’re seeing a drop of about 70 percent in the number of layout illustrators and stock images we need to book though. The main consequence of this, however, is that our creative directors are now having to deal with a much higher workload because they’re being presented with substantially more content to evaluate. Last year, we learned that copywriters and art directors who had a strong foundational understanding of their craft and a good grasp of their sector’s professional jargon produced much more impressive results with the help of AI than the ones who primarily relied on tutorials and trends for their work.

AI is the great disequalizer. How do you see this impacting less skilled workers? I’d be very worried if I was one of them. However, I’ve heard independent creators have only been seeing more commissions.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: