CEO of Microsoft AI, with a $100B budget.
"I think AI should best be understood as something like a new digital species."
and later
"AI is us. It's all of us."
Paraphrase: "To avoid existential risk, we should avoid:
1) Autonomy
2) Recursive self-improvement
3) Self-replication
We have a good 5 to 10 years before we will have to confront this."
His TED Talk is now online.
Tags: TED 2024 Mustafa Suleyman Microsoft Inflection AI
Such Mind Expanders. We had a dissection session, below...
Tags: Cephalopod TED 2024 Mind Expanders Dissection Session Squid
When the CEO of Humane, the AI pin company, "glitched" on stage (her words), she referred to her notes on her palm... just like the AI pin would have projected had she been trusting it on stage (detail below)
Tags: TED 2024 Humane AI CEO Bethany Bongiorno
At TED 2024:
"540M years ago, there were no eyes. Light went unseen. Trilobites were the first to sense light, and the Cambrian Explosion of body plans followed. Sight led to insight. Seeing led to understanding and action, the elements of intelligence. We are teaching machines to see."
"Seeing is for doing and learning. Spatial intelligence will catalyze a virtuous cycle of seeing/learning/doing."
Tags: Fei-Fei Li Stanford Professor AI TED 2024 Seeing ImageNet Evolution Machine Vision
His full slide deck: Plausible Futures (2035-2049):
#10 in deck: "we will have new food & fertilizers
We will have much better alternate protein production to replace traditional animal protein, and “green” fertilizer. And taste will far exceed that of traditional cows!"
The others:
#1: expertise will be free
We will be capable of having AI-based near free tutors for every child, and doctors for every citizen 24x7, accessibly & affordably
#2: labor will be near free
We will have a billion bipedal (and other) robots freeing humans from the slavery of the bottom 50% of undesirable jobs, creating a larger industry than the auto industry
#3: computer use will grow expansively
There will be a billion+ programmers all “programming” in natural, human language, dramatically increasing what can be done with computers. Computers will adapt to humans, not humans to computers, democratizing computers
#4: AI will play a large role in entertainment & design
Music and entertainment could be plentiful, and personalized for you and your mood! Content creation would be commoditized, but not change the celebrity-consumer relationship
#5: internet access will be mostly by agents
Most consumer access of the internet could be agents acting for consumers doing tasks and fending off marketers and bots. Tens of billions of agents, on the internet will be normal
#6: there will be clean, dispatchable electric power
By 2050, fusion boilers will retrofit and replace coal and natural gas boilers reducing the need to build whole new fusion plants. Superhot geothermal >400C is also a real alternative
#7: cars could be displaced in cities
We could replace the majority of cars in most cities with personal autonomous transit as on-demand, affordable, public transit with little public financing. This will reduce congestion, increase city street throughput
#8: flying will be faster
We will have Mach 5 planes that get us from NYC to London in 90 minutes – on sustainable aviation fuel, making the world closer!
#9: resources will be plentiful
We can discover more natural resources than we consume in the next few decades. We will prove all resource doomers on
lithium, cobalt, copper, etc wrong.
#11: from the practice to the science of medicine
We will be capable of providing precision care based on patients’ -omics, as well as AI models for each individual enabling N=1 personalized simulation, therapeutics, drug dosage
#12: carbon will have solutions, if we have time
Carbon emissions could be a smaller issue
because we will get entrepreneurs to develop &
scale better technologies for cement, steel,
agriculture, transportation, power production,
DAC, etc. Most such efforts will fail but enough
will succeed to solve the problem of carbon
emissions in the critical areas, well before 2050!
"Abundant, awesome, technology-based, “possible tomorrows” are likely if we allow them to happen
... things that might slow down predictions
● Incumbentresistance
● Politicians capitalizing on fear for personal / populist gain
● Tech failures or delays
● Financial market conditions may kill a good idea (fall in chasm of a bridge too far)
● Anti- tech sentiment (naysayers and tech haters)
● Luddites hijacking the would-be advocate
● A few bad AI-related outcomes that get sensationalized and taint the water
● Left field events
● Instigators/entrepreneurs may not show up to make it happen"
-----
I agree with most. I think #2 will be far off as it is coupled to physical manufacturing, distribution and installation (compare to solar panels - humanity can’t build and deploy them anywhere close to global need) and the development/training is relatively tough because of data poverty. 11 will take frustratingly long due to regulation slowing experimentation rate and embedded complexity of our biology. Same for 12 - regulation will slow, and physical adoption rate may make it too late (for similar reasons mentioned already for robots).
I agree strongly with #1, 4, 6 (not sure about geothermal though), 7, 9 (deep seabed mining alone solves this), 10 (e.g. The Better Meat Co.) and agree with #3 (if we presume a broadened definition of "programming" that does not feel anything like programming today) and 5.
8 might exist but will be a niche impact on humanity vs others.
Actually I can’t say for sure which of 2, 11 and 12 will be last. They will probably frustrate us by being 10 years farther off than the advocates would like. As I ponder further, I’d put 11 in last place for completion of the vision (progress is made every year, but a small % of total potential).
Tags: Vinod Khosla TED 2024 Predictions