A good overview but we need to move beyond these giant models into more efficient forms of AI that are less resource and energy hungry (lower carbon footprint) and can scale across the edge of the network- across devices and sensors. @TDataScience
@deeplearn007
-
Efficient AI: Lower Power, Edge Computing, Reduced Carbon Footprint
By
–
This is why we need to move towards more efficient forms of AI entailing:
Less computational resources;
consume less energy + lower carbon footprint;
Scale on the edge of the network (across devices & sensors) in power constrained environments;
Possess memory;
Are ultra low -
Symbolic Logic Integration in Knowledge Graphs and Neural Networks
By
–
There is active research on integrating symbolic logic into knowledge graphs / graph neural networks.
— AI (@DeepLearn007) 10 octobre 2023
My team and I did quite a bit of work on this 18 months ago integrating LLM models based on Transformers with Knowledge Graphs for logic and more explainable AI.
Here is a good… https://t.co/J7slPUa2k2There is active research on integrating symbolic logic into knowledge graphs / graph neural networks. My team and I did quite a bit of work on this 18 months ago integrating LLM models based on Transformers with Knowledge Graphs for logic and more explainable AI. Here is a good
-
Transformers Evolution: Shift Toward Efficient Smaller Models
By
–
@TDataScience it’s great to see so much interest in Transformers after they were being overlooked by many Data Scientists & publications for too long. I was writing about them and working with them 3 to 4 years ago. But now we need to move to more efficient smaller models that
-
Graphene technology faces production scaling challenges like quantum computing
By
–
Grapheme is cool & has awesome properties but like quantum computing we’ve been waiting for many years for it to take off. Like quantum computing there are many challenges to scale it get it to mass production in terms of quality as well as quantity. Perhaps, unlike quantum
-
Neuromorphic Computing Will Revolutionize AI Before Quantum
By
–
Look to Neuromorphic Computing (when not if) first. It will revolutionise AI way before Quantum Computing (if not when)
-
1300 Experts Sign Letter AI Force Good Not Threat
By
–
An open letter signed by more than 1,300 experts says AI is a "force for good, not a threat to humanity".
-
AI Regulation: Balancing Innovation Against Misuse
By
–
AI Regulation is on the way. Hopefully proportionate, reasonable & well thought-out to balance innovation vs misuse.
-
AI Job Creation Through Skills Training and Physical Automation
By
–
For avoidance of doubt in current state and likely state of next few years they are unlikely to displace millions of jobs. On the contrary with correct skills training they will help generate more jobs. Truly advanced AI that can impact our real-world with physical interactions
