Is Google done?

Is Google done?

Yesterday Google released Gemini, and apparently outperformed ChatGPT by a far stretch.

Getting some real competition in the LLM space would surely be nice. Having one dominant company outperforming everything that's out there is not a good thing in the long run, so we're welcoming some healthy competition here. However, Gemini is not that thing ...

Gemini is NOT on pair with ChatGPT

Unfortunately, Gemini is not on pair with ChatGPT version 4. First of all it looks like Google basically applied the oldest marketing trick in the world on us, by literally stacking rubbish on top of rubbish, and more or less flat out lying with its performance. The video you see where it's solving the 3 cups trick for instance, was created by prompt engineering it multiple times, manually giving it instructions, with examples of rights and wrongs. Of course all that prompt engineering was completely removed in the marketing video to create a "Wow effect".

I'm not going to provide you with references here, since all you need to do is to literally go to Google or YouTube while searching for "Google Gemini", and have Google (pun!) provide you with the truth about their own testing methodology, and facts about their own LLM model, but apparently it was all for shows.

Besides, the Gemini you've got access to now in Bard is not their most advanced version, and is actually more on pair with GPT3.5. Their best model won't be released before next year, at best. However, lawyers at Google have already raised concerns about Gemini, implying it won't even come out in January, but might be postponed to Q2 2024 - At which point the best math I've seen on the subject implies Google might have an LLM that's on pair with ChatGPT version 4 early in Q2 2024.

This of course ignores the fact that OpenAI will probably have GPT5 out the door before Gemini Ultra is even available for most people. Implying Gemini will be obsolete before it's even publicly available.

Is Google collapsing?

Google is being attacked from every possible angle you can imagine these days. LLMs such as ChatGPT for instance are 100% incompatible with their existing business model. You simply can't make money using ads on an LLM for reasons others have explained in more details before. Running an LLM is too expensive compared to a Google search query, so the math doesn't add up.

The Grim Reaper at Google's Door

YouTube, another one of Google's primary products is also struggling. TikTok has taken a huge share of their market, and there are signs showing us that YouTube is obviously struggling. A couple of months ago they started blocking ad blockers (the irony!) This is a sign of a business unit having difficulties resulting in that they've got to start "collecting pennies". If YouTube was actually profitable, they wouldn't care about ad blockers. Less than 2% of its users are using ad blockers, and losing 2% of ad viewers isn't something a profitable business unit would care about.

By blocking ad blockers the only thing they achieved was a second surge for ad blockers, resulting in that those 2% all of a sudden increased to 10% in a couple of months. Using my own YouTube experience as the primary example of how YouTube is not making money is easy for me. My own ads are in the following category:

  • 40% of my YouTube ads are in Greek
  • 30% of my ads are programming tutorials for noobs
  • The rest of my ads are about BTC schemes, "get rich fast" courses, personal growth websites, etc

To drive the above numbers into context, realise the following truths:

  • I don't understand Greek, so 40% of YouTube's ads I am not even cognitively capable of understanding. For all I know, they might be about some kick ass deals, but I wouldn't know because I don't understand a single word of Greek
  • I've got 41 years of software development experience, so websites teaching me "how to get started with software development" are literally of zero interest to me
  • I have never bought a BTC in my entire life, I don't trade, and I have never paid for an online course in my entire life

You would believe that a company basically having my entire life story would be able to provide me with better and more personalised ads, but apparently not. I suspect YouTube ads are the worst possible form of advertisement you can possibly choose - And in fact I've got experiences to prove it. I once spent $3,500 on YouTube marketing, and I got 2 leads. So far I've used $20 on Facebook ads in 2 days now, and we've already got more leads for my $20 worth of Facebook ads than I got from $3,500 on YouTube.

Conclusion

I don't believe Google will collapse, as in go chapter 11. But I do believe their party days are over. I realise they had all time high profits in 2023, but so did Microsoft in 2001. 5 years later Microsoft had completely lost its status as the largest company in the world, and I suspect Google is on its way down too the same way Microsoft fell from grace 20 years ago.

And my primary example of Google suffering is Google themselves. Go to Google and search for anything, literally anything. Notice how the entire SERP is literally stuffed with ads? You've got to scroll below the fold to find results that are not paid placements. This of course degregades their product, resulting in mass migration over time away from their platform, to alternatives such as Bing and DuckDuckGo ...

Google will probably still be around 5 years down the road, but their powers are slowly diminishing, and their relative value will probably never be able to match their current status. As for my personal feelings about this ...?

Jo Sundar, say hi to Yahoo on your way down there ... 😉

Thomas Hansen

Thomas Hansen I am the CTO of AINIRO.IO AS. I am a software developer with more than 25 years of experience. I write about Machine Learning, AI, and how to help organizations adopt said technologies. You can follow me on LinkedIn if you want to read more of what I write.

Published 8. Dec 2023