
Ezioath
@ezioath
0 reply
0 recast
0 reaction
26 replies
863 recasts
1105 reactions
1 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
Right now, weโre in the era of narrow AIโ, or even more specialized models that excel at specific tasks (think language generation, image recognition, or playing Go). AGI requires a leap: an AI that can reason across domains, learn from minimal data, and adapt to entirely new challenges without being spoon-fed instructions. Experts are split. Optimists like those at DeepMind or OpenAI hint at breakthroughs within a decadeโsay, by 2035โdriven by exponential compute power and algorithmic tweaks. Pessimists, or maybe realists, argue itโs still 50+ years off, pointing to how little we understand about human cognition itself.
The next big breakthrough? Itโs likely a cocktail of three things:
1. Better reasoning model
2. Embodied AI
3. Energy-efficient compute 1 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
1 reaction