How Google is Planning To Beat OpenAI

In April, Alphabet CEO Sundar Pichai took an unusual step: merging two large artificial intelligence teams -- with distinct cultures and code -- to catch up to and surpass OpenAI and other rivals. Now the test of that effort is coming, with hundreds of people scrambling to release a group of large machine-learning models -- one of the highest-stakes products the company has ever built -- this fall. The Information: The models, collectively known as Gemini, are expected to give Google the ability to build products its competitors can't, according to a person involved with Gemini's development. OpenAI's GPT-4 large-language model can understand and produce conversational text. Gemini will go beyond that, combining the text capabilities of LLMs like GPT-4 with the ability to create AI images based on a text description, similar to AI-image generators Midjourney and Stable Diffusion, this person said. Gemini's image capabilities haven't been previously reported.

Google employees have also discussed using Gemini to offer features like analyzing charts or creating graphics with text descriptions and controlling software using text or voice commands. Google is betting on Gemini to power services ranging from its Bard chatbot, which competes with OpenAI's ChatGPT, to enterprise apps like Google Docs and Slides. Google also wants to charge app developers for access to Gemini through its Google Cloud server-rental unit. Google Cloud currently sells access to more primitive Google-made AI models through a product called Vertex AI. Those new features could help Google catch up with Microsoft, which has raced ahead with new AI features for its Office 365 apps and has also been selling access to OpenAI's models to its app customers.

Yawn

[ihavesaxwithcollies]

Why would anyone care about monolithic corporations fighting each other? Here's an idea, break them up.

Google can't fight it's way out of a wet paper bag

[MIPSPro]

The post "Don't be Evil era" has come. They are nothing but a big boring mega-corp. The just failed the last 10 projects they took on. Their search engine is now completely focused on censorship, not giving decent results. Even fucking Yandex is better these days. They will be lucky to write an AI spellchecker.

Re:

[account_deleted]

Comment removed based on user account deletion

Re:

[pereOlthwaite]

It no longer seems to matter if whatever it is you're looking for has been out of copyright for over a century, Google will now actively confound your attempts to find it. Search functions on sites which host such material are provided by Google, or one of its numerous offshoots, so you still have to turn to Yandex to find what you're looking for even if you're right there standing in the archive already. But surely all will be forgiven, so long as they provide us with a shiny new way to fiddle about with b

NYTimes says 'NO' to its data used to train AI

[schwit1]

https://www.theverge.com/2023/... [theverge.com]

Re:

[ToasterMonkey]

https://www.theverge.com/2023/... [theverge.com]

No, what they're doing is saying is you can't have it for free. They're probably going to negotiate with companies like OpenAI and Google for access, which is not remotely the same thing as saying no.

Greed hinders them from progressing

[MindPrison]

You remember when Google actually was a knowledge database search engine?

Then greed took over, until the place was just one pile of advertisement junk, all fighting to be on top of that search pile of garbage results.
They could have fixed it overnight, but then they would lose out on trillions of dollars, the money is just too good. They knew what was right, but chose wrong, and they are losing to the competition without the bias of ads, just results - like Google was born out of.

R.i.p. Google.

Re:

[wakeboarder]

Its the natural lifecycle of companies, the real problem is trying to squeeze money out of customers for investors. Doctrow said it best: https://kottke.org/23/01/the-enshittification-lifecycle-of-online-platforms

Re:

[gweihir]

Indeed. Stopped using them a few years ago, they were just unusable for most things. These days Google is essentially a one-trick pony and that trick is getting more and more outdated.

I've heard this story somewhere

[phantomfive]

Sounds like they are going to try to bury OpenAI. [arstechnica.com] Did Sundar also throw chairs? Two CEOs who made their companies worse.

Google 3.0

[groobly]

Google 1.0 was the winning lttle engine that could search company.
Google 2.0 was the winning eat everyone's lunch advertising company.
Google 3.0 is the behomoth everything company, er, call it X, no, ok, Y. All that will happen, like any big company, is that 90% of the effort will be infighting over dominance inside the company.

Whatever it is they plan and do...

[LordHighExecutioner]

...this [killedbygoogle.com] is the destination for the solution.

Underwhelming

[jvkjvk]

So, their big plan, according to the summary, is to essentially create a pipeline between a ChatGPT type submodel and a Midjourney type submodel, and have them talk, so the interface can display images to you as well as text.

Wow. Talk about innovation!

Wonder how happy those teams are

[wakeboarder]

if they don't like working the same way.

OpenAI caught them off guard

[oumuamua]

To its credit, Google made AI a company priority years ago well before it was trendy. They invented transformers; the key to todays LLMs. They just got blindsided by ChatGPT thinking they had more time than they did. Their paper count dwarfs other companies:

In 2020, Google had 178 papers accepted and published at NeurIPS, while Microsft had 95, DeepMind had 59, Facebook had 58 and IBM had 38. Amazon had less than 30. For the same year at ICML, Google had 114 papers accepted and published, while DeepMind had 51, Microsoft had 49, Facebook had 34, IBM had 19, and Amazon had 18.

https://www.cnbc.com/2021/01/2... [cnbc.com]

Re:

[MrPaul]

The problem with your argument is that Google is a business, not a research university. It measures success by introducing products that generate revenue, not publishing papers in academic journals.

They are panicked

[gweihir]

And they cannot catch up. What OpenAI has is not very good, but it probably took about a decade to build. That is a process that cannot be really accelerated. All Google can deliver is something much, much worse and even with ChatGPT the problems are pretty obvious with the staggering nonsense it often delivers. What Google will have will just be crap and no hiding it.

Google looks _old_ these days. Old and creaky and obsolete.

In other words, do what's easy, not what matters

[gestalt_n_pepper]

> Gemini will go beyond that, combining the text capabilities of LLMs like GPT-4 with the ability to create AI images based on a text description

But they won't:

1) Reduce hallucinations

2) Implement iterative, self-correcting rule based reasoning in a neural net architecture to train the Chatbot piece.

Looks like management has already prepped this to be a disappointment.

The necessity for such an ostentative top-down a

[blugalf]

dministrative measure ... is living proof that Google has effectively calcified into a Microsoft. Sustained by the inertia of solid bread-and-butter business, but no longer nimble enough to produce true innovation at core while still trying to.

Understand their purpose

[SkyLineDP]

I would like to understand their purpose. In the end, this seems to be done for even more control of users. This trend raises privacy concerns.