Tech3 | A year of plateau and new directions for AI; Digital transformation deals drive IT recovery; and more

Tech3 | A year of plateau and new directions for AI; Digital transformation deals drive IT recovery; and more

One quick thing: Magicpin enters into quick commerce for food delivery 

In today’s newsletter:

  • A year of plateau and new directions for AI
  • Digital transformation deals drive IT recovery
  • Aye Finance files draft papers for IPO

Event alert: You're invited to Moneycontrol's India Fintech Conclave. We have a star-studded lineup featuring policymakers, bankers and top founders who are reshaping the fintech landscape in India. 

  • It's happening tomorrow in Mumbai. RSVP here.

A year of plateau and new directions for AI

As 2024 draws to a close, the AI landscape finds itself at a crossroads. 

From GPT-3.5 to GPT-4, the tech world was abuzz with excitement. But as we close the chapter on this groundbreaking year, a new reality emerges: the AI landscape is hitting a plateau.

The scaling limits: For years, the mantra of the AI industry was "bigger is better." 

  • By throwing more data and computing power at AI models, researchers believed they could unlock superhuman intelligence 

However, recent developments suggest that this approach is reaching its limits.

  • OpenAI, at the forefront of this scaling race, is now rethinking its strategy

  • Ilya Sutskever, a key figure in the AI community, acknowledges that the era of scaling is waning

  • Sutskever believes that the focus is shifting towards a new era of "wonder and discovery"

The data dilemma: As AI models grow, so does the need for high-quality, human-made data. 

  • The internet alone can no longer supply the diverse, specialised datasets that are needed to train the next generation of AI systems 

This has led companies like OpenAI to form partnerships with publishers and other organisations to acquire data—an expensive and time-consuming process.

Then comes the cost: Training cutting-edge AI models is an incredibly expensive endeavour. 

  • The cost of developing models like GPT-4 and Google's Gemini runs into the millions, if not billions, of dollars. As models grow larger, so do the costs

Looking to 2025: In the years ahead, AI will likely evolve away from simple model scaling and toward more nuanced forms of growth. 

  • OpenAI has already shifted its focus from just building larger models to creating AI “agents”—intelligent systems that can take on more complex, real-world tasks

Google’s Sundar Pichai has already hinted that the low-hanging fruit in AI research has been plucked, and the future will demand deeper, more profound innovations.

As we say goodbye to 2024, one thing is clear: the AI race is far from over. 

Go deeper

Picture credit: Microsoft Copilot

This is a short version of the MCTech3 newsletter. Sign up here to get the full edition in your inbox every evening on weekdays. 

Harshad Bansal

Product | Growth Analytics & Insights | HT | Paytm | Digital | News | Content | Subscriptions

6d

Nice well written roundup of the genAI journey

Like
Reply

To view or add a comment, sign in

Insights from the community

Others also viewed

Explore topics