|
Getting your Trinity Audio player ready...
|
When OpenAI launched ChatGPT in November 2022, it created a storm in the tech industry. It was able to set records in gaining the fastest million users at the time. ChatGPT was built on Transformer architecture, which was created and later made open-source by Google.
However, when ChatGPT gained traction, a narrative formed: “Google is dead; ChatGPT killed it.”
Content creators across platforms started creating content hyping ChatGPT, setting this narrative. Google grew concerned that ChatGPT might eventually replace Google Search. People were prompting ChatGPT more than they were searching on Google. Meanwhile, Google’s search results had started losing trust, often topped with sponsored content that prioritized ad revenue over quality.
ChatGPT, by contrast, was pulling users away and, with them, their attention. It was definitely making the Google investors nervous.
Google launched Bard in March 2023, their ChatGPT competitor. Bard was buggy and failed to strike a chord with most users. Its name lacked impact, and it wasn’t close to ChatGPT in terms of performance.
Google later rebranded Bard to Gemini and launched its flagship models in February 2024. Gemini performed on par with, and in some benchmarks even better than, GPT-4. However.
The Gemini Image Generation Controversy
When Gemini was released, it also had the image generation capabilities.
Some users reported that Google’s Gemini AI was generating historically inaccurate images, including racially diverse Nazi soldiers and other bizarre results. It triggered major backlash. Experts called it a failure of basic safety checks. Others criticized Google for over-correcting diversity in ways that defied historical context. The controversy got so big that some even called for the resignation of the CEO, Sundar Pichai.
In fact, Google parent Alphabet reportedly lost some $90 billion in market value in the same month of Gemini’s release due to this controversy.
However, the company went into stealth mode and started correcting their mistakes, one after the other.
Google releases NotebookLM
NotebookLM, built by Google and powered by Gemini, lets you upload documents, PDFs, notes, and research papers and ask questions directly from them.
It’s like having a research assistant that truly understands your material and context.
It wasn’t just another chatbot. It became a powerful tool for writers, students, and researchers who deal with a lot of information. Instead of wasting hours digging through pages of notes or dense research papers, you could simply upload your files and ask questions like, “What does the author say about this topic?”
And in seconds, you’d get a clear, context-aware summary, not a generic response, but one grounded in the actual content you provided. It turned information overload into something manageable, making research faster, learning easier, and workflows way more efficient.
I uploaded Google’s prompt engineering guide into NotebookLM, and tried the “Audio Overview” feature — it generated a podcast-style audio summary, which you can listen to below.
It’s a game-changer for understanding complex topics. Super useful for students or anyone who prefers learning by listening. Apart from that, you can chat about it, asking questions, and it will answer you based on the document that was uploaded. In addition to that, you can create a mind map based on the document.
Honestly, if there’s one underrated AI product that deserved more attention in 2024, it was NotebookLM.
Gemini 2.5 Series exceeds AI Benchmarks
In December 2024, Google launched Gemini 2.0, its next-gen AI model family, starting with the experimental Gemini 2.0 Flash. It was a big step into true multimodal AI — capable of handling text, images, audio, and video, putting it in direct competition with OpenAI’s GPT-4o. Gemini 2.0 Flash also stood out for being one of the most affordable models for developers and quickly found its way into several Google products.
By March 2025, Google leveled up again with Gemini 2.5 Pro Experimental, its most powerful model yet. It offers a massive context window, strong multimodal support, and advanced reasoning.
Google backed its claims with benchmarks showing Gemini outperforming other leading LLMs, and early hands-on testing seems to confirm that. This version doesn’t just generate — it reasons, with built-in self-checks baked into how it thinks.
I’ve been using Gemini 2.5 Pro at work, and it’s hands down the best LLM I’ve worked with. It’s solved some really challenging problems I was stuck on, and even when it didn’t give a direct answer, it pointed me in the right direction.
That alone saved me hours.
As of this writing, Gemini 2.5 Pro leads the Artificial Analysis Intelligence Index chart. Google made this model free for everyone. Yes, free.
Just a few days ago, in April 2025, Google announced the release of Gemini 2.5 Flash in preview via the Gemini API, available through Google AI Studio and Vertex AI.
Gemini 2.5 Flash is designed for speed and efficiency while still retaining reasoning capabilities, making it a solid choice for developers who want performance without sacrificing intelligence. The model has a better performance-to-cost ratio compared to its competitors. It’s incredibly fascinating to see how fast Google has been releasing its models to compete and stay ahead in this AI race.
Closing Thoughts
How did Google pull this off? Weren’t they supposed to fall behind in the AI race?
Well, they didn’t give up. They took the hits, owned their mistakes, and kept at it. They even (allegedly) paid top dollars to retain top AI talent, even those who weren’t actively contributing, just to keep them from going to competitors.
The CEO, Pichai, has been actively running polls on 𝕏, and engaging with people on the platform to get feedback to improve, something rare for a company this size.
I’m pretty sure that, at the pace Google is releasing products, they’ll stay ahead, even with strong competitors like OpenAI, Anthropic, and Microsoft in the race. This is what happens when a trillion-dollar organization works like a startup.
If this article provided you with value, please support me by buying me a coffee—only if you can afford it. Thank you!
