DailyPost 2384
CHATGPT TO GPT-4 TO ARTIFICIAL GENERAL INTELLIGENCE (AGI)
From all projections in the long history of AI, this discussion was slated to happen around a decade from now. It was also believed that it the meantime lots of ethical issue with regards to this technology could get sorted out. The history of AI form 1950s has followed the cyclical pattern of AI inflection point and AI winter and in this manner till around 2010 when AI caught research. IBM Watson going commercial in 2010 and DeepMind getting founded* in the same year have been major milestones. AI victories in games like Jeopardy, Go and Chess have been paving way for something tectonic to happen.
The AI Inflection Point that ensued was promised to say. The datafication of mankind, readily / hugely available storage and immense computing power unleashed was making this possible. Added to that was the magic sauce of machine learning; the algorithms of human transformation. With the tech wherewithal in place and the urge to make it happen helped take some monumental steps, but that was not enough. The market and understanding of AI were not there and it would take ages for that to happen, at the pace it was moving
Though taken over by Google in 2014 DeepMind was not able to make waves in the wide world. With the intention to transform the landscape of AI research, not for profit OpenAI was created in 2015 with a clear-cut mandate, to create transformative products, where neither the industry nor the academia /research could make any earth-shattering headway. The journey of OpenAI for the last eight years has been a AI journey which was not expected to be traversed so fast. It made its mark with GPT and its progress has been more than satisfactory. The release of GPT-3 released 2020 was certainly a milestone, but the talk of Artificial General Intelligence, AGI, was still not in the air. It could not be seen in the foreseeable future.
Over 300 next generation applications are being generated powered by GPT-3. Then came the GPT-3.5 powered ChatGPT, on the dialogue mode. It was released on 30th of Nov 2022, a conversational intelligence, took the world by storm in not time. The numbers tell it all. The biggest tech advancement in recent times. Everyone and everything are getting impacted. Then came GPT-4 which was released on 13th March 2023. A large multimodal model accepts image and text inputs, emits text outputs. ”While less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.” The trajectory of AI today is clearer than ever before. With reinforcement learning coupled with large scale human feedback, Artificial General Intelligence i.e., human level intelligence may be a just be a few years away.
ARTIFICIAL GENERAL INTELLIGENCE (AGI) IS LOOMING LARGE ON THE HORIZON.
Sanjay Sahay