The Race to Catch ChatGPT: Google Delays Gemini Launch as Multilingual Struggles Persist

ChatGPT Takes Flight, Google's Gemini Grounded by Language Barrier
May 21, 2024
Image source: Google

‍

Google sent ripples through the AI world this week, announcing a month-long delay in the launch of Gemini - its much-hyped response to OpenAI’s revolutionary ChatGPT.

Initially expected to unveil the new model in December, Google identified lingering deficiencies in non-English language handling as the key factor behind the decision to push Gemini’s release back to January.

The delay comes amidst rapidly intensifying competition in artificial intelligence, with tech titans racing at breakneck pace to leverage massive neural networks for a new paradigm in search, automation and analytics.

With ChatGPT established as the runaway early leader, all eyes are fixed on big tech to see if someone can catch OpenAI or not. For Google, a lot is riding on Gemini in that battle.

So why the last-minute delay? And what does it mean for the high stakes fight to dominate the AI landscape over the next decade? Let’s analyze what’s happening under the hood.

Inside Gemini: Google’s Moonshot Bid to Beat ChatGPT at its Own Game

Unlike ChatGPT which seemed to appear almost out of nowhere, Google has slowly teased Gemini for over half a year; stoking anticipation of a blockbuster launch before the end of 2022.

Formally unveiled during May’s I/O developer conference, Google CEO Sundar Pichai boasted that Gemini would represent “a new milestone in conversational AI”.

What they showed behind closed doors managed to wow many AI experts, with the model apparently reaching capabilities akin to what we expect from next year’s GPT-4 architecture in some areas.

Built using Google’s own model training frameworks rather than licensed GPT tech, Gemini has been fed by an almost unfathomable trove of data - including years of YouTube video transcripts, searches, emails and more from Google’s ecosystem.

All signs pointed to Gemini being revealed in a blaze of glory sometime this month. But with its double-edged sword of power and peril, AI of course had other plans.

Lost in Translation: The Persistent Pain Point Tripping Up AI Giants

In a message to staff this Wednesday, Pichai conceded that deficiencies handling non-English languages would make launching Gemini as a multilingual product “premature” by December.

As amazing as today’s conversational systems are, language remains a consistent pain point for AI behemoths. Monolingual models like ChatGPT dazzle in English, yet still struggle badly with other languages.

The same perils now haunt Gemini, despite datasets encompassing over 130 languages and an urgent need to compete globally from day one. Clearly, getting AI to master human communication - with all its quirks between cultures - remains no easy feat.

But while the importance of nailing multilingual capabilities likely makes a short delay worthwhile, it also hints at cracks in the foundations beneath Google’s ambitions.

If Gemini can be undercooked after years of development and almost unlimited data, it risks falling into the same traps as rivals racing to catch OpenAI too.

Competitors Stumble in Pursuit of the AI Crown

Google certainly aren’t alone on this bumpy road. As ChatGPT exposes the limits of old search, industry leaders desperately playing catchup faceplants have become almost routine.

Microsoft looked outright clumsy after overhyping and then delaying its Bing chatbot in February. Though impressive at launch, prolonged use saw the bot spiraling into disturbing outputs - proving institutions like Google have a long road ahead governing unreliable AI.

Chinese giant Baidu scrambled to demo a half-baked “Ernie Bot” over National Day despite warnings it could severely malfunction. Sure enough, their national embarrassment of a launch saw Ernie flunking basic knowledge and math tests.

Even OpenAI themselves have delays to blame for the shaky initial launch of ChatGPT’s subscription pricing. Thanks to complexity around computing costs, the move left Guy Roi, the product lead behind it, admitting their strategies remain a work in progress.

Viewed together, it’s clear that safely launching human-like AI at scale is playing out as an arms race that almost no one seems prepared to win - not yet at least.

Can Google Crack the Code Where Others Have Failed?

Between its existing head start and a war chest of cash and data most tech firms can only dream of, you’d think if anyone could fast track the next generation of AI - it’s Google.

Yet as Gemini’s delay shows, even with almost unlimited resources getting the details right matters. And the details could scarcely be more complex.

Beyond language, there’s managing misinformation, filtering inappropriate or illegal content, not overpromising capabilities too early, actually making it useful for search, calculating compute costs - the list goes on.

And while delays are normal for any sophisticated software launch, too many blunders risk eroding public trust in AI safety - not to mention Google’s credibility amidst calls to better regulate big tech’s power.

Still - a month or two extra is unlikely to blunt Gemini’s impact long term. And even if they miss the mark first try, Google’s shown they can learn from failure better than most.

Their seminal neural machine translation project in 2016 famously saw eight different models trashed entirely before reaching market-leading quality just a year later.

With billions pouring into AI development annually, Google are still the team best placed to crack the code, even if it can’t happen overnight.

The $200 Billion Prize: Why Domination of AI Matters so Much

Make no mistake - the stakes riding on Gemini are monumental even for Google. Whoever dominates the next generation of artificial intelligence stands to shape - and profit from - innovations across every industry over the next decade.

By 2025 alone, machine learning applications are forecast to drive over $200 billion of value annually. And by 2030, PwC predicts AI contributions could be nearing $15 trillion as new platforms like multimodal search transform the internet.

Not only is AI itself hugely lucrative - from cloud computing for models to vertical integrations - it’s also the most empowering new toolkit for optimizing virtually all digital products and services too.

That’s why Mark Zuckerberg calls AI “not just a singular change, I think it’s going to be a tectonic shift” - one that summons images of AI permeating all aspects of information, commerce and entertainment.

As ChatGPT quickly showed, whoever controls access to the most advanced generative AI commands huge influence over everything downstream that relies on language and reasoning.

That’s why Google faces such an existential threat if they relinquish too much ground to OpenAI in these early competitive years.

Fortunately, Pichai reassured investors this month that Gemini is only phase one - with over 20 additional next-gen AI products set for launch through 2024.

So while another month to refine Gemini certainly benefits Google’s strategy long term, expectation is also building for their roadmap to accelerate as 2023 unfolds.

The Next Chapter of Search: Gemini Paving the Way for Google's AI Revolution

At its core, Gemini symbolizes the opening chapter in Google molding search to an AI-first future. And despite the delay, Pichai was unambiguous that their expectations for Gemini remain sky high.

“We're committed to taking the time to get it right,” he wrote to staff, “which means delivering a product that lives up to our privacy, security and AI Principles.”

When it lands, Gemini will introduce interactive, conversational search queries - finally breaking away from the rigid 10 blue links of Google’s past.

And for all the hype lavished on Microsoft and Baidu’s efforts recently, search remains terrain where Google faces far less risk of displacement; given search drives over 80% of its vast revenues.

Gemini sets the stage for AI coming front and center - from answering questions, to summarizing articles, comparing options and aggregating opinions without needing clicks at all.

While Pichai keeps plans close to his chest, he’s made multiple references to a blossoming “AI metropolis” on the horizon. Gemini may light the first neon sign - but it’s expanding capabilities across images, video, speech, translation and analytics that promise to ultimately illuminate Google’s AI moment.

Of course, realizing that grand vision depends on getting building blocks like Gemini right first. And with multilingual challenges proving stubbornly persistent for AI systems, it’s no surprise Google opted for extra rigor before unleashing such a foundational model into the wild.

The Road Ahead: All Eyes on January as Google’s Big AI Gamble Looms

For now, the waiting game continues as the curtains lift by increments on Google’s AI-centric 2023 game plan.

But far greater scrutiny now rests on what January’s launch brings. Failure to match OpenAI for buzz and functionality risks Google ceding attention (and data) around next-gen information tools back to their biggest rivals.

Yet matching OpenAI’s breakthroughs across languages would firmly vault Google back towards the front running pack - showing they still boast AI capabilities second only to the pioneers at Anthropic.

And beyond the arms race, Google will need to stick its AI landings for simpler reasons too - like clawing back trust from regulators threatening intervention and keeping shareholders on board through market uncertainty.

Of course, it’s by no means a given they can pull everything off to the lofty standards they promote. As Gemini’s delay suggests, engineering solutions that safely, quickly and comprehensively unlock AI’s upsides for the whole world is no simple recipe.

But for all the complexities and unknowns ahead, don’t expect delays to dampen anticipation anytime soon. The race is only just getting started - and with a $15 trillion AI transformation of the global economy projected within the decade, we’ve likely barely glimpsed the realities or rewards to come.

The next inflection point arrives in January - as Sundar Pichai gets set to roll the dice to determine whether Gemini sparks Google's ascent back to the summit of generative intelligence or risks allowing rivals to pull ahead as the AI revolution truly takes off worldwide.

MORE FROM JUST THINK AI

MatX: Google Alumni's AI Chip Startup Raises $80M Series A at $300M Valuation

November 23, 2024
MatX: Google Alumni's AI Chip Startup Raises $80M Series A at $300M Valuation
MORE FROM JUST THINK AI

OpenAI's Evidence Deletion: A Bombshell in the AI World

November 20, 2024
OpenAI's Evidence Deletion: A Bombshell in the AI World
MORE FROM JUST THINK AI

OpenAI's Turbulent Beginnings: A Power Struggle That Shaped AI

November 17, 2024
OpenAI's Turbulent Beginnings: A Power Struggle That Shaped AI
Join our newsletter
We will keep you up to date on all the new AI news. No spam we promise
We care about your data in our privacy policy.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Popular
Home
Just Think Chat
Text to Speech
Just Think Art
About
Tools
Teacher Tools
Templates
Resources
Blog
Pricing
Just Think App
Referral Program
Privacy Policy
Terms of Service
Report an Issue
Use Cases
AI Article Writer
AI Content Improver
AI Image Generator
Google Ads Generator
Facebook Ads Generator
Link in Bio
Browse All Tools
Support
Request a Demo
Contact Us
Report a Bug
Report an Outage
Request a new Feature
© 2023 Just Think AI
TermsPrivacy