5 Major Shifts In AI Landscape Since Sam Altman’s Last Visit To India

Sam Altman is in the headlines for his visit to India. Last time, he came to India in June 2023. At the time, Altman interacted with the media and other stakeholders from the business and technology sector, where he commented on the possible potential of Artificial Intelligence (AI) affecting jobs, formulation of the rule related to AI and pervasiveness of human intelligence over AI. 

Since Altman’s visit to India in 2023, the global AI landscape has evolved significantly, bringing transformative changes that are increasingly shaping people’s daily lives. 

Let’s take a look at the changes that have emerged since his last visit to India.

AI models become cheaper to develop 

When Altman on his 2023 visit to India asked whether a small team of startups with $10 million can create a model like ChatGPT, he said, “It’s totally hopeless to compete with us on training foundation models. His comment also received huge backlash from different sectors of India. 

Though OpenAI has not completely disclosed the exact cost, ChatGPT developed by Altman’s company took approximately $100 million to $1 billion. While developers of DeepSeek claimed that they built an AI model under a $600 million budget. The launch of DeepSeek by a Chinese startup suggests that AI models can be developed at a lower cost, demonstrating a more affordable approach to building advanced AI systems. However, there is still a debate going on the internet that DeepSeek’s hardware spending alone was well in excess of $500 million, as per media reports. 

Multimodality In AI Models

Initially, AI models were limited to providing responses in text form. Now, in order to enable more interactive and versatile experience, they evolved to assist users through images, audio, and video. 

Several AI models launched after 2023 included the feature of multimodality like Gemini by Google, GPT-40 by OpenAI and DALL-E 3. Multimodality feature of AI models has enabled users to perform diverse activities plus also reduced hallucination and bias results. 

Agentic AI Giving Real Time Information 

Unlike generative AI, which relies on user prompts to provide answers, images, or other outputs, agentic AI goes a step further to make decisions and execute them. It gets the work done. 

Industry experts say 2025 will be the year of Agentic AI. A McKinsey report last year described “agentic” systems as digital systems that can independently interact in a dynamic world. The management consulting firm believes that AI agents eventually could act as skilled virtual coworkers, working with humans in a seamless and natural manner. 

More Resources, Greater Capability? Not necessarily 

Scaling laws is the principle that refers to the addition of computational resources and eventually performance of AI models changes as an outcome. These laws offer insights into effectively scaling AI systems to enhance their capabilities and efficiency.

Launch of OpenAI's GPT-2 and GPT-3 models are some of the key examples of scaling law; these advanced versions of GPT are much more efficient than their predecessors. 

Push for Sovereign AI 

With AI set to drive every major decision, a nation’s ability to develop and control its own AI infrastructure is becoming increasingly critical for staying competitive and securing its future.

Sovereign AI ensures that nations can dictate their own rules, preserve their cultural identity, and address their unique needs. It is the key to staying ahead, both economically and geopolitically, in an era where tech-driven power defines the global balance.

Sovereign AI encompasses physical and data infrastructures. The latter includes sovereign foundation models, such as LLMs (large language models), developed by local teams and trained on local datasets. While countries can take control of their future and create models that work for their benefit, ethical frameworks are crucial to ensure fairness, especially for nations with limited resources or talent. These issues will be key going forward.