Page 5 - AI Vol 1: Foundations of AI

LLMs DO NOT JUST EXECUTE
PROGRAMMED TASKS;
THEY CAN UNDERSTAND
AND GENERATE HUMAN-LIKE
RESPONSES AND MAY ONE-DAY
SOON BE ABLE TO USE THESE
ABILITIES TO INTERACT WITH THE
PHYSICAL WORLD.

Unlike the AI/ML technologies that we are already
familiar with, which often focus on specific tasks
or utilize only certain aspects of AI, the current learned from vast data sets, enabling them to
wave of generative AI technologies, including handle a wide array of queries and tasks, many of
LLMs, are advanced systems designed not just which may not have been explicitly anticipated
for narrow applications but as versatile, adaptable by their developers. LLMs do not just execute
platforms capable of a wide range of functions. programmed tasks; they can understand and
A key distinction of LLMs, compared to earlier generate human-like responses and may one day
AI forms, lies in their emergent properties, such soon be able to use these abilities to interact with

as the ability to understand context, engage in the physical world.
complex conversations, and even reason in a way
that at least appears similar to humans. This shift from task-specific AI to more general
and adaptable systems like LLMs marks a
While previous technologies, like voice assistants, significant evolution in the field. It is a transition
use AI for specific functions like speech from machines that simply execute predefined
recognition or natural language processing,

they ultimately still rely on specific, rule-based
programming to execute tasks. For instance, IT IS IMPORTANT TO
when a voice assistant like Google Assistant UNDERSTAND THAT WHAT SETS
uses AI to recognize the command “turn off the LLMs APART IS NOT JUST THEIR
lights,” the actual action of turning off the lights ADVANCED CAPABILITIES BUT
is still managed by traditional programming that ALSO THEIR POTENTIAL TO
interacts with smart home systems. In contrast, REDEFINE OUR INTERACTION
LLMs are not programmed in the traditional sense. WITH MACHINES AND THEIR
While they may operate within certain guidelines ROLE IN OUR PROFESSIONAL

or guardrails, their responses and actions are AND PERSONAL LIVES.
not pre-defined by explicit rules. Instead, they
generate responses based on patterns they have

FOUNDATIONS OF AI | LOZANOSMITH.COM VOLUME 1 | 5