Every week, my team at Invector Labs publishes a newsletter to track the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:
From the Editor: Before AGI We Will Have Narrow AGI
Artificial general intelligence(AGI) has been the ultimate dream of AI researchers since the early days of the space. AI systems that can surpass human intelligence across different tasks has proven to be elusive to the current generation of AI technologies but we are one to two breakthroughs enough from making it a reality. However, I’ve recently been thinking that the first generation of AGI systems might not be as generally intelligence as we dream.
AI systems today have proven to be efficient at mastering very narrow tasks on specific domains. From a technology evolution standpoint, it is conceivably that those systems would achieve general intelligence within their specific domain before expanding to other domains. AI agents that are building knowledge of medical pathologies have a better chance to master all medical knowledge before becoming experts in video games. From that perspective, we are not going to go from narrow AI to AGI but rather to narrow AGI.
Now let’s take a look at the core developments in AI research and technology this week:
Facebook AI Research(FAIR) published an insightful blog post about new research breakthroughs in natural language processing that would be included in the next generation of NLP applications like digital assistants.
IBM Research published a research exploring methods to improve the robustness of neural networks to adversarial attacks.
Google published a paper unveiling the research behind Project Euphonia, a speech recognition system for people with disabilities.
Cool AI Tech Releases
DeepMind open sourced bsuite, a toolkit to evaluate the behavior of reinforcement learning agents.
Uber’s Chief Scientist gave an insightful interview about how the transportation company applies AI at scale to complex problems.
LinkedIn unveiled their work on Data Hub, a metadata search and discovery platform optimized for data science workloads.
AI in the Real World
During an earnings report call, Nvidia CEO Jensen Huang described AI as the most powerful force of our times.
Alphabet’s subsidiary DeepMind has been at the forefront of some of the biggest AI breakthroughs of the last few years but they are also loosing over $500 million a year. Are they on the right track?
Data labeling startup raised $12 million to streamline data annotation.