Present and future of AI Flashcards
Why brain inspired computing
AI needs to be more and more human-like
-Devices become smaller and smaller
-We need to cut down on energy consumption
-This is a powerful way of dealing with massive amounts of complex data
Current applications
Self-driving cars
AI in smartphones
AI in healthcare
Large language models (LLMs)
Launched in 2022
Every question costs 5 cents in computer power
Can be used by other companies for specific applications
Generative AI
AI capable of generating text, images, movies, etc - using generative models
Learn the pattern and structure of their input data and generate new data that has similar characteristics
E.g - chatgpt, copilot, bard, DALL-E, midjourney
Artificial general intelligence (AGI)
AI that surpasses human capabilities in the majority of tasks
Also referred to as strong AI, full AI, human level AI
Superintelligence
Superintellignece or singularity is reached when AI becomes as creative and smart as humans and can further develop itself into AI that is more intelligent than humans.
Law of accelerating returns (Ray Kurzweil 1999)
Development of technology accelerates
Better technology allows for faster development of even more advanced technology etc
This applies especially to AI and leads to an exponential increases in the intelligence of systems
Chatgpt can already write computer code
Transhumanism
-The idea of merging with technology
-The transformation of the human condition by applying sophisticated technologies which are able to greatly modify or enhance human intellect and physiology.
Transhumanism examples
Emotion recognition - detection of drowsiness of drivers, allows robots to respond to human emotion
- Wearables - augmented reality, health monitoring
-Internet of things - connecting smart devices and using information detected by sensors of other devices to respond as an integrated system
-Smart pills and micro-implants = act as intra-body sensors
-Smart prosthetics - prosthetics with enhanced functionality
-Brain computer interfaces - implanted chips to record and stimulate brain activity.
Baby x
Training in a real world setting =
object recognition
speech recognition
Language processing
Emulating humanlike behaviour
-Detailed physiological model
-Affective computing/emotion AI; emulating stress system
Spider
Applications:
Natural interaction with digital agents/bots
Artificial companions for elderly patients with neurodegenerative disorders.
Smart Prosthetics BCI and AI
Restoring the link of brain-muscle control:
-Brain encodes complex signals to control muscles in the upper arm
- Muscle movement is detected using ultrasound
- AI helps to interpret ultrasound data into correct movements of the fingers
- Similar applications are now tried for vision
AI concerns
Navigating the next industrial revolution
-AI is a disruptive force in society = AI taking over our jobs
- Increasingly difficult to tell truth from fiction
- Privacy concerns - usage during the paris olympics
-AI can develop undesired strategies for the goals its trained for
- Misalignment in social media apps
AI in warfare - autonomous weapons
Bill gates and elon musk ask for regulation of AI as for nuclear technology
- Autonomous weapons can be used on the battle field with no risk for the attacker
-Do we want an AI to decide about life and death