Is the AI Hype Cycle already almost over?

From time to time I hear things about how the current wave of AI hype, driven by LLMs and GenAI, is just a fad. Another "AI Winter" is on the horizon! We're on the precipice of another "trough of disillusionment"!

Source: The Gartner Hype Cycle on Wikipedia

Here's why I don't think that is the case.

1️⃣ General Accessibility - AI is not new, but accessible AI for the average consumer is. When you look at the history of major technological advancements, it's typically the point at which it becomes accessible to the average person that it reaches critical mass and becomes a staple of our lives - and the "tail" of sustained growth usually lasts for decades. Great comparisons include the computer, microwave oven, and GPS.

2️⃣ Dedicated Hardware - Pay attention to what Groq has achieved in building a new type of chip that is specifically optimized to run LLMs. Now you hear big players like Microsoft talking about the same "TPUs" on stage at their grand unveilings as well. We use GPUs today because they're the most powerful chip we have, but they were built and optimized for gaming and 3d rendering - NOT AI. New hardware takes a long time to build and refine but when it comes en masse, it will completely change the game and make real-time voice/image/text processing commonplace. Even if we don't see a single further breakthrough in generative AI, this will still change your everyday life.

3️⃣ On-Device Use Cases - Thanks to the above, every tech giant is now moving towards the promise of on-device AI, whether it be for the iPhone (Apple), Android (Google), PC (Microsoft), or others. This is the biggest opportunity for LLMs to be integrated into daily life, but most people have not actually seen or used anything like this before. Even devices that are ideal for this, like Google Home/Alexa, still haven't been upgraded with this capability.

4️⃣ Business Adoption - Lastly, I have personally seen time and time again how we are only scratching the surface of the possibilities of this technology for business use. The most common uses I see for LLMs are still just summarization tasks and single zero-shot prompts, and the vast majority of companies are still not using LLMs in any official capacity. This will change, but it will take time for it to become clear in the market that the companies using this technology to its potential are actually achieving an advantage over the competition. Almost every startup I see in the space is trying to "sell a shovel", but precious few are actually digging into the real opportunities below the surface and in the niches. 💎

---

This content was originally posted on LinkedIn, and we invite you to join the conversation there