As AI becomes ordinary, we must understand the presumptions we are encoding
“It’s a truism in the AI world that as soon as a technology becomes reliable, it’s no longer considered to be AI. Machine translation used to be the most interesting…
The forward momentum of AI is probably far too powerful to restrain or direct
“The past few years have seen a distinct decline in the trust that citizens have in their institutions – political, business, educational, etc. Fake news and skepticism about science, expertise…
The fight to gain first-mover and network effects advantages is on. Will AI meet expectations?
“Of course, many people are thinking about the issues around AIs – especially the major industry players – but I’m not confident that values beyond efficiency, novelty and profit will…
The problems raised by AI cannot be solved simply by bolting guardrails onto existing systems
“Popular visions of created intelligence as a utopic or dystopic force date back more than two centuries. Today it is possible to envision that artificial machine intelligence could cause dramatic…
We will be better off overall in 2040 if general purpose AI does not progress much, though social recalibration will be disruptive
“I start with two assumptions. The first is that we won’t see much progress on general-purpose AI in the next 15 years or so. If this is wrong, all bets…
‘We need to allow room for human discretion and struggle,’ important parts of being human
“My biggest concern at the moment is that we are trying to rein in AI before clearly defining its boundaries. “In the spring of 2023 the White House put out…
AI seems magical now, but it really offers only flawed and limited promise
“While we will continue to see significant advances from AI in many ways, the raw power of simulating intelligent behavior through LLMs will plateau as a result of model collapse…
Humanity must take action to assure individuals’ rights, dignity and superiority over AI
“AI is not a weekend experiment, it has some of the most inquiring minds behind it and some of the coolest talents. “The first thing we must do is trust…
Outside of job losses, narrow AI will be mostly beneficial; this may change when AGI arrives in 3 to 15 years
“AI timelines are very difficult to predict. My own prediction, as an expert in neuroscience, neural networks and cognitive architectures, is that we can expect to advance to having self-improving…
Human-machine rules should achieve the reality we want for our children and grandchildren
“Intelligence cannot be artificial, so ‘artificial intelligence’ – isn’t. The idea of more-complex computational machinery begs two questions: Who is going to use it? And in what ways? The real…