Galaxy S24 AI Features: You Shouldn’t Be All Happy About It

We all are happy because of the new Samsung Galaxy S24 new AI features that help to better have the best experience with our smartphone, but we should not yet get excited to we know these thing.

Samsung isn’t the first company to dabble in the AI craze sparked by apps like ChatGPT and Midjourney. The two industry leaders in smartphone silicon, Qualcomm and MediaTek, recently put on a spectacular display to highlight the generative AI capabilities of their most current flagship and mid-range CPUs.

Things have changed. The newest buzzword in phone marketing is going to be AI. What precisely are we in for, though? It’s still unclear how these alleged “AI advancements” will bring any real benefit. Or perhaps it’s just an old trick that needs to be repackaged as a native app or go by a different name.

Let’s begin with the phones in the Galaxy S24 series. The Snapdragon 8 Gen 3 chip from Qualcomm will be included in the package. The chip manufacturer is asserting quite a bit about the generative AI powers of its latest flagship chipset. For instance, it is claimed to use Stable Diffusion AI technology to generate a picture in less than a second when given text instructions.

The explanation provided by Qualcomm for how it enhanced Stable Diffusion technology for on-device use is rather sophisticated. Rival MediaTek also asserts that Stable Diffusion, an on-device AI feature, would enable text-to-image creation in less than a second for its flagship Dimensity 9300.

We currently own a phone that runs on Qualcomm’s most recent flagship. The iQoo 12 is the device in question, and it’s noteworthy to note that the phone’s marketing materials make no mention of any generative AI tricks—especially the ones that Qualcomm and MediaTek are pushing.

What you ought to do with these photos of ninja cats created by AI

Assuming Samsung is the company that will provide such text-to-image creation capability. In the end, what is it going to accomplish? We are still determining whether Samsung will include the text-to-image technique in one of its applications or will be packaged within a third-party program.

The actual question is how much value it will offer to our regular smartphone usage habits. What will a typical purchaser of a Galaxy S24 do with the visuals produced by a single text prompt? They may use those AI-generated pictures to spice up conversations or create a stir on social media.

However, there’s still some conflict here. These AI graphics must be created from a line of text, saved locally, copied straight to the clipboard, and then pasted into your chosen chat program. The best outcome is if Samsung directly included the image-to-text generating technique in the keyboard.

Once more, stickers, GIFs, and emoticons may accomplish the same goal as going to all the bother. Furthermore, these AI-generated graphics’ 512 x 512-pixel output is an insufficient resolution for a job presentation or college project.

Moreover, it is unlikely that the system will be free. A Premium option is mentioned in the MediaTek presentation video. Customers purchasing Galaxy S24s may find that they have reached their text-to-image token limit after a certain number of outputs; at that point, they will either be required to pay a membership charge or be demoted to a lesser speed tier for picture generation.

However, if that turns out to be the case, the entire discussion could be more worthwhile because there are already workarounds available, such as OpenAI’s Dall-E. You may pay for ChatGPT Plus to take advantage of quicker and more detailed picture production with the most recent Dall-E 3 model, or you can use it to produce photographs for free. Numerous text-to-image generators exist, and that’s just one of them.

The Snapdragon 8 Gen 3 is the “first to support multimodal Gen AI models,” according to Qualcomm. This implies that in addition to voice input, the chatbot—based on Meta’s Llama model—will also accept text and image inputs when it runs natively on the phone. That’s not unique, either, as ChatGPT-4 already provides that feature, albeit behind a paywall.

Is it indispensable for us to have these AI features?

The AI image expansion feature is one of Qualcomm’s most exciting new features. In essence, you may stretch the canvas of a picture in any direction you like, and the generative AI on your smartphone will creatively create new items within the original frame by intelligently generating pixels based on your text prompt.

Seeing an image grow as additional things are added and the horizon widens like a skilled painter touching up his work is an incredible trick. However, you want to do something other than that to photos you take with your phone and store in memory. Expanding your memories using generative AI is akin to contaminating them with fictitious images and situations you have never encountered.

Another high-risk strategy for deploying generative AI on phones is summarization. It works well for reading news articles and staying current on recent field advancements. However, this method would only be noticeable if there is very little friction. Consumers may summarise a news story without opening a separate app on the same browser page.

Why not switch to an app that already does it if the latter is the case? For instance, Kevin Systrom, a co-founder of Instagram, created the beautifully designed app Artifact, which uses AI to compile content for you.

Some websites and applications like Inshorts already provide news in the form of condensed chunks. Shortwave is an excellent software for your inbox that does more than summarize email chains for you at no additional cost.

Voice-based photo editing is only one of the intriguing tricks that on-device generative AI can do. It sounds advantageous, but it isn’t easy to envision the sheer convenience it will bring when one-tap filters and fine-grained sliders provide an even faster and more satisfying flow for editing media on phones.

Let’s now discuss the practical application of AI to initiate general conversations or to find answers that would otherwise necessitate internet-based research. We are going to encounter qualitative issues once more.

Because resources are readily available, generative AI models that run natively on phones, such as Meta’s Llama, are not the most sophisticated of their kind. Google is the only place to look. Only Google’s Gemini Nano, the smallest of its large language models, is supported by the Pixel 8 Pro. Why not use a more robust language model and go straight to something like ChatGPT or Pi via their respective mobile apps?

Where is the need for generative AI now?

Generative AI is now at its finest when we talk about an alternative provision of the best layer of security, severing smartphone functions from the cloud, and the need for constant internet access. However, these on-device AI tips must also function as an assistant, like Siri, Alexa, or Google Assistant.

Better still, they ought to integrate with the helper. Give your generative AI assistant instructions to gather all the cat pictures in your collection, put them together into a collage, and email them to your father. You can even ask it to locate you the cheapest ticket for the upcoming weekend, make the ideal schedule for a day trip to Disneyland, and organize all of that data nicely on Google Calendar.

Furthermore, there is less reason to be concerned about data privacy if an on-device generative AI tool maintains all operations local to your smartphone and no longer transfers your data to cloud servers. That is, at least. I am trying to understand what Samsung means by the Galaxy AI vision it is currently marketing. Still, seeing if it can deliver meaningful generative AI experiences or just a series of gimmicky, barely functional tricks will be interesting.