Universal Do you think that GPT/LLM is the future of App Store search?

[ivanicin]

To say that App Store search is broken isn't anything particularly new.

At this point Apple appears to be using only the app title, subtitle, keywords, category name and some metrics that it has regarding the apps that may influence the app to rank higher for searches but not to rank for more searches.

Not to say that Google Play is much better, but still it does use some more sources like app description, user reviews, web links, text from the pages that link to app... While it is far from good it is at least in some aspects "less worse" - at least it can handle more complex queries and return somewhat reasonable results.

If I had to describe current state it would be this - you can only search for some fairly broad terms and you will get either most popular apps in that category or the apps that hyper target those keywords, and usually considering the number of apps competing they have to be both.

Today I came to idea to ask Chat GPT on some very specific solution that I might need and see if it can respond. For example, I have used something that I am familiar with - PDF text to speech reader that skips headers and footers. This is something that many people would need, but if you search in the App Store you will get no results and you will have to revert just to PDF text to speech reader which would return thousands of results where 99% of them don't have this frequently needed feature.

And guess what? Chat GPT answered.

It came to me that this would hugely improve the landscape of iOS apps in many ways:

• people would get the apps they need and not the apps with biggest marketing budget that in many cases aren't what they need
• it would be nearly impossible to trick this as it uses so many sources on such a complex way that it would reduce title and subtitle spamming that is a standard now in the App Store - even multinational companies do that
• this would help more developers to thrive as everyone could have his own niche based on the features - not just few popular apps would get all the money because they have the biggest budget to pay for ads and be first both in advertised and organic searches...

I understand that the last one may be the problem for Apple - ads are probably now bigger revenue generator than the commission it gets from devs so it will be hard to run away from that system. Still as GPT/LLM might be a paid service in iOS in the future, it would be reasonable at least for those users to have GPT/LLM powered search in the App Store.

What do you think?

 

[NEPOBABY]

Searching can always be better but it's the presentation of results that matter. If you think search results aren't good enough, those are presentation and design decisions.

An LLM consumes a lot of energy compared to simply querying databases and indexes. If everyone in the world was doing that we would run out of energy pretty soon, energy costs would spiral out of control, rents would climb a lot, businesses would shut or inflate prices, bosses at energy firms would take home huge bonuses and laugh at how poor everyone becomes.

 

[ivanicin]

Searching can always be better but it's the presentation of results that matter. If you think search results aren't good enough, those are presentation and design decisions.

An LLM consumes a lot of energy compared to simply querying databases and indexes. If everyone in the world was doing that we would run out of energy pretty soon, energy costs would spiral out of control, rents would climb a lot, businesses would shut or inflate prices, bosses at energy firms would take home huge bonuses and laugh at how poor everyone becomes.

Well if you can't search for the feature you need and you need to rely on Chat GPT (or even Google), I don't see what presentation and design has to do with that?

Regarding the energy, if there is a good candidate for GPT it is App Store as results are highly valuable and searches are highly infrequent...

 

[NEPOBABY]

Regarding the energy, if there is a good candidate for GPT it is App Store as results are highly valuable and searches are highly infrequent...

Searches are already algorithmic. They have been since inception. There's no need to make them so intensive that they consume masses of energy. Bing is already doing this and users are complaining that their LLM is producing garbage.

AI should not e too pervasive, not only because we will all die from energy poverty, but also because you will end up with AIs scraping and training on AIs. That's a downhill slope towards poorer quality data.

Reliable qualified people need to be in the loop at all times, not just in creating tools we use but also curating and creating content and information.

GPT worship is highly misplaced, delusional, and no different from believing that sky gods are going to save us.

 

[ivanicin]

Searches are already algorithmic. They have been since inception. There's no need to make them so intensive that they consume masses of energy. Bing is already doing this and users are complaining that their LLM is producing garbage.

AI should not e too pervasive, not only because we will all die from energy poverty, but also because you will end up with AIs scraping and training on AIs. That's a downhill slope towards poorer quality data.

Reliable qualified people need to be in the loop at all times, not just in creating tools we use but also curating and creating content and information.

GPT worship is highly misplaced, delusional, and no different from believing that sky gods are going to save us.

OK, I apologize for hurting your religion feelings by discussing of possiblity to include GPT algorithm in the App Store. I have deep understanding for religious people, but I am definitely not into discussing of their beliefs or in this case angers.

 

[pacorob]

I think the main problem with the current App Store is that Apple seems to almost fully focus on app promotion for apps that have 1) paid them money to promote their app 2)have in-app purchases and therefore include it their lists instead of looking into the quality of an app and creating collections as users do in the AppRaven community.

Until that time that community (and other sites such as TouchArcade, AppApp.io, Reddit iOS, iOSgaming communities) are the better place to look for a certain type of app. Either ask in the community (by posting) or use the detailed search or search in the collections on e.g. genre (e.g. Apple Watch Music, Apple Watch Health, calendar, RTS. RPG, Platform), type (e.g. controller support)

 

[ivanicin]

I think the main problem with the current App Store is that Apple seems to almost fully focus on app promotion for apps that have 1) paid them money to promote their app 2)have in-app purchases and therefore include it their lists instead of looking into the quality of an app and creating collections as users do in the AppRaven community.

Until that time that community (and other sites such as TouchArcade, AppApp.io, Reddit iOS, iOSgaming communities) are the better place to look for a certain type of app. Either ask in the community (by posting) or use the detailed search or search in the collections on e.g. genre (e.g. Apple Watch Music, Apple Watch Health, calendar, RTS. RPG, Platform), type (e.g. controller support)

It isn’t a simple thing - apps with bigger marketing budgets get better metrics and it is hard for App Store to ignore those metrics.

I am more concerned with very low semantic surface that Apple takes into the account when forming the database for keyword searches and their matches.

Other than that for me it is reasonable for Apple to put the app that has 100k ratings over the app that you like and that may have just 100 ratings… Just it is not reasonable to push it for literally any search in the category even if the app doesn’t work well for that.

 

[pacorob]

It isn’t a simple thing - apps with bigger marketing budgets get better metrics and it is hard for App Store to ignore those metrics.

Other than that for me it is reasonable for Apple to put the app that has 100k ratings over the app that you like and that may have just 100 ratings… Just it is not reasonable to push it for literally any search in the category even if the app doesn’t work well for that.

Apple could still decide (they do already earn lots of money from those bigger developers with major budgets) to feature indie apps in certain list to get them more out there and help these smaller developers create more exposure as I've read from indie developers that this really helps for people to find their apps. There are so many great apps in the App Store or there have been also great ones where the developer unfortunately no longer invest in them since the don't get enough exposure to maintain a business model for the developer.

As far as the ratings are concerned I really have my doubts about those. There are quite a few fake reviews in the App Store especially for these horrible ad filled freemium game apps that you see advertised both on e.g. YouTube ads and in the App Store. See also:

Fake app reviews impact developers more than you think (June 2023)

Apple App Store and Google Play flooded with fake reviews (March 2023)

Towards understanding and detecting fake reviews in app stores (2019, scientific article)​