Google PaLM 2 and AI will change games, for better or worse

Announced at Google I/O, the PaLM 2 is one of the latest models to join the rapidly growing AI trend. However – there is one important aspect that sets this one apart, and it will change gameplay forever.

The successor to the original PaLM large language model has four sub-paradigms: Gecko, Otter, Bison, and Unicorn. The smallest model – Gecko – is aimed at mobile and mobile devices, which will work on the device and even offline. While the OpenAI language model is being implemented on a mass industrial scale (take Microsoft and IBM, for example), PaLM 2’s Gecko mini-paradigm will turn that dynamic on its head.

Already appearing across Google Workspace in Docs and Gmail, the API is openly accessible to developers and tinkerers alike. However, we’re not sure if we’re excited or worried that the gaming industry will be the next batch to adopt.

The gaming industry is already dependent on artificial intelligence

Credit: Bethesda

In late March 2023, Ubisoft announced Ghostwriter, an AI writing tool designed to help game writers bark (filler phrases and hums adopted by NPCs during cutscenes and the like). This Ubisoft PR report asserts that the AI ​​service “does not replace the video game writer” but will instead assist with some menial tasks.

Public reaction to newly introduced AI automation software such as Ghostwriter has been mixed. The dangers of substituting jobs and distilling creativity into a single, homogeneous line of code are two of the biggest tactics of opposing AI – though Ghostwriter itself doesn’t seem so innocuous – in its current state at least.

Role-playing games are limited in scope by the amount of dialogue (quality) and script writing that the team of creators can muster. This year we saw the release of the Fallout 4 AI dialogue mod that adds 300 new lines to the game. There is no doubt that this model took a long time to create and develop, and there was certainly a sense of organization and editing that the modders oversaw to ensure that some of the notorious and illogical sentences of artificial intelligence were not seen ending in a wasteland. However, Google PaLM 2 can completely eliminate this need for the human touch.

Why PaLM 2 could be the AI ​​that changes games forever, or ruins them

Google’s PaLM 2 technical report publishes some fairly promising results for smallest sub-paradigm performance, and while Gecko won’t really step on the toes of what ChatGPT is used for, it might be looking to offer new solutions instead.

Being able to run from a local device, even when offline, Gecko will be able to generate in-game dialogue and paths almost instantly for games ranging from mobile devices, handheld game consoles and PCs, without having to go back to a remote server somewhere across the globe. . For reference, if a game wanted to take advantage of AI-generated content, it would need to contact a remote language model and wait for a response. However, the seamless integration of unique content that is generated on the device could be the second biggest advancement in gaming since 3D graphics – and PaLM 2 may be the first readily available language model to achieve this.

Of course, the quality of content generated by Gecko will be limited. Content generated offline on an Android device will not be compared to content analyzed by thousands of graphics cards, but it will lay a solid foundation for local AI generation.

An AI-generated image of a 2D game world.
Credit: Night Cafe “2D Sprite Style, Overworld Map”

For repetitive gacha games and dungeon crawlers – this is it could Be cool. Imagine endlessly generated maps, characters, enemies, and weapons, along with matching dialogue that neatly complements them. A whole new kind of infinite game may be awakening, in which each world and narrative is woven according to what the player wants – but is this a cause for concern rather than optimism?

As evidenced by this attempt to enjoy a DnD game run by ChatGPT Dungeon Master, the results aren’t particularly promising. The chatbot failed to understand the creative spark the DM needed, even though it “seemed to be watching the same magic trick over and over.”

If AI-generated games really take off – we’d envision a solid period of mediocre, unconvincing releases featuring soulless, derivative content. Google Bard, for example, had a very poor launch, after constantly posting plagiarized content while contradicting itself as much as possible. Games generated by artificial intelligence will have equally difficult launches. AI models need frequent training, and while PaLM 2 is promising, there is still a feeling that we are not quite at the point where it is ready to go. Open AI Jason Weifor example, he believes that artificial intelligence has produced novels about games along with “[compelling virtual worlds] Designed by the language models themselves” is less than two years away, despite the current explosive rate of development, there is no telling how much shorter this time frame will actually be.

While games could certainly benefit from the ease of generative item descriptions and NPC dialogue here and there, the eventual consolidation of all content creation emanating from a single digital identity is inevitable – and Google’s PaLM 2’s frighteningly small scale may prompt us to come close to this.

User-customized AI games are on their way.

Cover art created with Dall-E 2, then edited in-house.

#Google #PaLM #change #games #worse

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top