IDS2935 Siri is my Superpower Unit 4

HMC

Human Machine Communication

HRI

Human-Robot Interaction

HCI

Human Computer Interaction

HAI

human centered artificial intelligence

machine

device and people's communication with it is part of a much bigger phenomenon

human machine

relationship between humans, machines, and cultural dimensions that are often overlooked

communication

serves as disciplinary marker

communication process is often defined as

a human only process

ex- transaction model of communication

mass communication replaced interpersonal with

depiction of single source media organizations transmitting information to human audiences

technology is designed to function as

things with which people interact or carry out tasks

ex- mobile phone

apps are learning your preferences and adapting to complete tasks more efficiently

in HMC, technology is conceptualized as more than a channel or medium

enters the role of a communicator

HMC changes the communication model

interdisciplinary nature of HMC

engineering connects communication and technology

communication with machines has never been far from human context

Man-Machine Communication

two way convo that is goal oriented and aimed at specific objective

NASA definition of HMC

human instructs a machine, interacts during execution, and accepts information from it

Cybernetics

the science of communications and automatic systems in machines and living things

Cybernetics definition of communcation

process of information transfer among people, animals, and machines

Shannon/Weaver Communication

process between humans and machines

McLuhan Medium Theory

focuses on implications of form and function of media

social shifts

studies media as communicators

Carey's take on technology

tech is cultural, it represents how the world works AND coerces world into working in terms of representation

NLP

natural language processing

What is NLP

allows people to interact with tech as if it were human

Communicative AI

tech that performs tasks with some level of human intelligence

functional dimension

how AI tech is designed as communicators and how people perceive them within this role

communication with a machine is close to mass communication

Relational Dynamics

how people understood AI in relation to themselves and themselves in relation to AI

networked self

ways in which people leverage these technologies to reimagine who they are

Metaphysical Implications

Communicative AI is designed to function in an even more human like role than early computers

Blurred line between tech and human communication

Why study social robots? To challenge our...

notions of medium and media

understanding of the communication partner

notions of boundaries of communication

Social robots challenge our notions of medium and media

They aren't just transmission channels

Transcend the role of a medium

Ex- CASA paradigm

Social robots challenge our understanding of communication partner

Seen as "communicative other"

Communication partner is no longer human

HMC is not advocating for abandoning human in communication research

Social robots challenge our notions of the boundaries of communication

They vary in morphology

Communication with social robots may be different from FTF and not really comparable with it

FTF fallacy

Belief that face to face communication is the gold standard

Bing

AI chat feature available for long talks about almost anything

Bing split personality

search engine assistant

or

Sydney- dark side

AI hallucination

AI chatbots aren't sentient

they predict next set of words and make up facts and emotions

ChatGPT

Generative pretrained transformer

generates sophisticated text

tech doesn't go beyond the surface

ChatGPT may be the end of writing as a

gatekeeper

metric for intelligence

teachable skill

gatekeeper

three categories of intelligence for high schoolers

ChatGPT keeps them in the highest group

metric for intelligence

possible end of using writing as a benchmark for aptitude and intelligence

teachable skill

ChatGPT can fix a students essay

human editor found to be of higher quality

wordsmith

story writer

shimon

improv jazz player

the painting fool

algorithmic painter

Chess and AI

chess is considered a litmus test for intelligence

minimax algorithm

recursive and can deal with games like tictactoe and chess

deepblue

was able to play at the ability of a chess grandmaster

alpha beta pruning technique

stops evaluating more when at least one possibility has been found that is worse than one of previously evaluated moves

AlphaGo

designed to formulate its own set of instruction and to act on these "decisions"

designers have no idea what moves it will produce

random text generation

given a particular set of words, it is possible to generate a sequence that would be considered intelligible and meaningful without knowing what is being produced

issues with prefabricated templates

production of brittle and static algorithms

text feels artificial, mechanical, or "canned"

Quill

works with structured numeric and/ or quasi number data where nature and structure of data are predetermined by conventions of domain or domain experts

Quill steps

ingest data

derive features

propose angles

filter angles

locate "points"

render language

automated insights wordsmith

ingest data then analyzes the material to identify trends, changes, and other pertinent patterns

reassembly rules

specification for construction of new text by means of recombination of old and possible addition of new constituents

David Copes Emmy

analyzed existing musical compositions

rearranged the base components to generate a new original score

inner workings of Emmy

where each note in the composition would be represented by an event list (5 numbers) that could then be stored in data base and processed

Emmy processing steps

pattern matching

segmentation

recombination

output

automated collage generation

text retrieval and analysis

image retrieval and manipulation

scene reconstruction and rendering

Google DeepDream uses "inceptionism"

employing the company's image recognition algorithm but reversing its operations

generative adversarial network (GAN)

a form of unsupervised machine learning that employs two interconnected neural networks

generator

produces various candidates trying to emulate particular data distribution like an image

discriminator

evaluates different candidates that have been generated in order to differentiate between true data distribution (actual image) and various candidates produced by the generator

Semitic and conceptual problems with designing systems to be creative

creativity is difficult to identify or define

computer generated (CG) art

artwork that results from some computer program being left to run by itself with zero interference from a human artist