Page 6 - AI Vol 1: Foundations of AI

instructions to those capable of “thinking” and
“learning” in a way that is much closer to (and
may one day exceed) human cognition. As we 1956

explore the implications of these technologies, it Dartmouth Summer
Research Project the term
is important to understand that what sets LLMs “artificial intelligence” 1970
was coined.
apart is not just their advanced capabilities but Advancement in
rule-based ai systems
also their potential to redefine our interaction designed to mimic the
decision-making process
with machines and their role in our professional of human experts in
specific domains.
and personal lives. This evolution also brings
-
new considerations and challenges, which will ‘90s
‘00s
be critical to explore as we delve deeper into the Shift from rule-based
systems to deep learning
implications of these advanced AI systems. systems.

2011
02 B. HISTORY OF AI Apple released Siri, a
voice-powered personal
assistant that can
generate responses and
take actions in response
In the mid-20th century, a revolutionary idea 2014 to voice requests.
began to take shape: what if machines could think Facebook developed the
deep learning facial
like humans? The symbolic kickoff of this era recognition system 2016
DeepFace.
was the Dartmouth Summer Research Project on AlphaGo defeats the
world’s top players of “Go”
Artificial Intelligence in 1956, organized by John a board game known for
its deep strategic
McCarthy, a computer scientist who coined the 2017 complexity.
Transformer AI model
phrase “artificial intelligence,” along with others. introduced, built on
attention mechanisms.
The main objective of the conference was outlined 2018
in the original proposal: “The study is to proceed OpenAI released GPT
on the basis of the conjecture that every aspect 2019 paving the way for
subsequent LLMs.
of learning or any other feature of intelligence Google AI and Langon
Medical Center’s deep
can in principle be so precisely described that a learning algorithm
outperformed radiologists
machine can be made to simulate it. An attempt in detecting potential 2021
lung cancers.
will be made to find how to make machines use OpenAI introduced
Dall-E multimodal AI
system that can generate
language, form abstractions and concepts, solve images from text prompts.
the kinds of problems now reserved for humans,
and improve themselves.” Although this group 2023
Open AI announced
had ambitious goals, the challenges of mimicking GPT-4 multimodal LLM
that receives both text
human intelligence were more complex than and image prompts.
they or the technology available at the time were
able to overcome. Still, this conference laid the

foundation for future research and collaboration,

6 | VOLUME 1 FOUNDATIONS OF AI | LOZANOSMITH.COM