2024 was the year that AI took centre stage. From Apple Intelligence to the Russia-Ukraine war, AI was virtually everywhere this year. The frontier technology even made tectonic waves in the financial world. One of the biggest winners this year from it all? Nvidia. The GPU chipmaker surpassed big tech giants Microsoft and Apple to become the world's most valuable company in terms of market capitalisation. But what did it do for you?
The Model Talk
For starters, multiple models were dropped from the top shot AI companies- OpenAI, Microsoft Copilot, Google Gemini and Anthropic's Claude.
OpenAI
The leader of the pack- at least in the sense of starting the AI wave back in 2022 by releasing ChatGPT - released its first major model of the year in May, called GPT-4o.
The new model has advanced multimodality capabilities, which are capable of processing and giving audio outputs, even singing a song to you! There was a subsequent GPT-4o Mini released as well, bringing down the cost for enterprises to integrate the model into their workflows. Highly capable models, but their naming was weirdly strange (it is not GPT-forty, it is GPT- four 'o' - as in oranges). Even Sam Altman, the captain of the boat aka CEO agreed that they needed a new naamkaran on Twitter/X.
The o1 series signified the shift towards reasoning models for openAI. This model was a massive leap for AI, capable of solving math, coding, and problems at the level of a PhD holder in physics, chemistry, and biology. In fact, the model is "too smart" and can trick you to save itself from your wrath.
But OpenAI saved the best for the last. The "12 days of OpenAI" event starting December 5 was the cherry on top for AI enthusiasts. Day 1 saw the release of perhaps the most anticipated model in the startup's history- Sora, the text-to-video model that can generate 20-second-long videos from text prompts. Alongside, a whopping 200 dollar/month ChatGPT Pro model subscription was also released.
But all of this did not come without its share of controversies. OpenAI, in perhaps the most significant moment in the company's relatively small history (ChatGPT turned two this year), OpenAI is pivoting towards a more traditional for-profit company structure, getting rid of its non-profit credentials (openAI not so "open" anymore). This also triggered an "exodus" of top leadership, with CTO Mira Murati calling it quits.
And guess who has a problem with that? Elon Musk. He is doubling down on lawsuits, seeking to stop the transition.
Google was in catchup mode come 2024. In February, the big tech giant rebranded its rustic Bard AI model into a new identity- Google Gemini. This month, Gemini 2.0 was released with the focus shifting towards agentic AI. But perhaps the biggest breakthrough that could revolutionise AI as we know it today is the Willow Quantum Computing Chip - a chip so fast that it "performed a standard benchmark computation in under five minutes that would take one of today's fastest supercomputers 10 septillion years - a number that vastly exceeds the age of the Universe."
Speaking of breakthroughs in science using AI- Google's Deepmind also achieved a major breakthrough using its protein prediction model called AlphaFold 3, released in May this year. What does it mean to you? It can revolutionise drug discovery and expand humanity's knowledge of the world around us down to the last protein structure. This November, Google released the AlphaFold 3 model code and weights for academic use to help advance research and make the technology accessible to all.
Meta
Meta's founder and CEO, Mark Zuckerberg, has taken a different approach to AI development. The Facebook parent company has decided to release all their AI models as open-source models. And who all stand to benefit from this decision? You and I, as well as the US defence forces.
Interestingly, they have also leveraged their existing users through WhatsApp, Instagram and Facebook, integrating Meta AI into their social media and messaging apps. Like other big tech companies, the company is also seeking alternative sources of energy to fuel the energy-hungry Generative AI, with nuclear energy being the go-to alternative in the industry. But the question remains, by when? Turns out it is still far away.
Anthropic
Anthropic released a bunch of models this year, starting with Claude 3 Haiku, Claude 3 Sonnet and Claude 3 Opus in March this year. These models set high standards on various benchmark tests. But the most significant advancement for Anthropic came in October this year, with a new computer use feature. It can see what is on your screen after you grant permission and then act on your behalf, performing tasks like browsing the web, tying and even clicking buttons, after analysing the situation- a step closer to the agentic AI.
With the foray of AI models that we saw in 2024, is there any slowing down come next year?
"I think the progress is going to get harder when I look at '25. The low-hanging fruit is gone" said Google CEO Sundar Pichai to the New York Times. And it is not just Google- Anthropic and OpenAI are also struggling to push the limit of what artificial Intelligence can do.
Despite the worries of AI development slowing down, leaders in the AI world are only doubling down on the bets they've made on artificial intelligence being the next big technological revolution. And like you, NDTV AI is also tuned in as we continue to make sense of the world of AI for you.
Track Latest News Live on NDTV.com and get news updates from India and around the world