Click on a word or part of a phrase to learn more.

Scroll through the captions and click to skip to a caption.

Play video to start
Every day, it seems,
Every day it seems
there's a new, bewildering or frightening story
there's a new bewildering or frightening story
about AI in the news - how it's going to steal our jobs,
about AI in the news how it's going to steal our jobs
spread internet fakery on a colossal scale
spread internet fakery on a colossal scale
and generally take over the world.
and generally take over the world
But what exactly is AI - artificial intelligence -
But what exactly is AI artificial intelligence
and are the scare stories even true?
and are the scare stories even true
So, the first thing to know is that AI has been around
So the first thing to know is that AI has been around
a lot longer than you might think.
a lot longer than you might think
Its roots lie in an idea known as an "artificial neural network"
Its roots lie in an idea known as an artificial neural network
from the 1940s.
from the 1940s
A neural network is a bit like a team of interconnected workers
A neural network is a bit like a team of interconnected workers
that learn to solve problems.
that learn to solve problems
Each time they come up with a possible solution, it's marked.
Each time they come up with a possible solution it's marked
If there is room for improvement,
If there is room for improvement
they adjust and change their connections.
they adjust and change their connections
Over time the network becomes more efficient.
Over time the network becomes more efficient
And technology powered by neural networks is all around us right now.
And technology powered by neural networks is all around us right now
It suggests movies and music we might like.
It suggests movies and music we might like
It recognises faces and objects when taking photos on smartphones,
It recognises faces and objects when taking photos on smartphones
enabling features like facial recognition.
enabling features like facial recognition
It's heavily used by social media platforms to personalise our feeds.
It's heavily used by social media platforms to personalise our feeds
More recently, a form of AI known as "generative AI"
More recently a form of AI known as generative AI
is powering applications that can seemingly create new data.
is powering applications that can seemingly create new data
It can also power chatbots like Open AI's ChatGPT and Google's Bard,
It can also power chatbots like Open AI's ChatGPT and Google's Bard
which give humanlike responses to questions.
which give humanlike responses to questions
These are getting better at interacting with us,
These are getting better at interacting with us
and seemingly more humanlike.
and seemingly more humanlike
This can seem scary, but it's worth knowing point two.
This can seem scary but it's worth knowing point two
If you ask ChatGPT a question like this one -
If you ask ChatGPT a question like this one
"Why should we be concerned about AI?" -
Why should we be concerned about AI
it does a pretty good job of providing a response
it does a pretty good job of providing a response
that appears logical.
that appears logical
And with that convincing humanlike response,
And with that convincing humanlike response
it's easy for us to believe it understands what it's saying,
it's easy for us to believe it understands what it's saying
that it has feelings and motivations.
that it has feelings and motivations
It's understandable that we do this,
It's understandable that we do this
but it's worth remembering, right now AI can't think or feel,
but it's worth remembering right now AI can't think or feel
can't love or hate.
can't love or hate
ChatGPT and its counterparts
ChatGPT and its counterparts
are sophisticated sentence completion apps
are sophisticated sentence completion apps
that analyse our patterns of communication
that analyse our patterns of communication
and provide responses similar to the way humans would typically reply.
and provide responses similar to the way humans would typically reply
A bit like a "talking" parrot.
A bit like a talking parrot
Which leads us to point number three.
Which leads us to point number three
Chatbots can have an awkward relationship with the truth,
Chatbots can have an awkward relationship with the truth
technically known as "AI hallucinations".
technically known as AI hallucinations
You could also describe it as "making stuff up".
You could also describe it as making stuff up
The core of the technology is a model that uses probability
The core of the technology is a model that uses probability
to predict the next word, sentence or paragraph.
to predict the next word sentence or paragraph
It can generate seemingly plausible replies, but lacks the ability
It can generate seemingly plausible replies but lacks the ability
to assess truthfulness or the accuracy of its responses.
to assess truthfulness or the accuracy of its responses
So anyone thinking of using chatbots to write content
So anyone thinking of using chatbots to write content
needs to be careful
needs to be careful
they're not incorporating credible-sounding BS
they're not incorporating credible-sounding BS
that can be easily spotted by someone
that can be easily spotted by someone
who's actually done the research.
who's actually done the research
The idea of a racist machine might seem far-fetched.
The idea of a racist machine might seem far-fetched
But if AI is trained on data that's racist, biased or hateful,
But if AI is trained on data that's racist biased or hateful
then its output will be too.
then its output will be too
As we all know, racism, bias and hateful content can be found
As we all know racism bias and hateful content can be found
in abundance online.
in abundance online
In 2016, Microsoft launched an experimental chatbot called Tay,
In 2016 Microsoft launched an experimental chatbot called Tay
but quickly pulled the plug
but quickly pulled the plug
after it made racist and offensive remarks.
after it made racist and offensive remarks
It learnt to do this from interacting with users
It learnt to do this from interacting with users
on social media.
on social media
Microsoft apologised
Microsoft apologised
and promised to implement improved safety features in future.
and promised to implement improved safety features in future
This is why the ethical framework that governs any AI application
This is why the ethical framework that governs any AI application
is incredibly important, and why many are calling for safeguards
is incredibly important and why many are calling for safeguards
to prevent bias and hate speech to be built into AI systems.
to prevent bias and hate speech to be built into AI systems
For all the notes of caution, it can be easy to forget
For all the notes of caution it can be easy to forget
there's many potential benefits to AI.
there's many potential benefits to AI
It's set to truly revolutionise healthcare.
It's set to truly revolutionise healthcare
AI has already discovered new drugs
AI has already discovered new drugs
and is being used to identify cancer cells
and is being used to identify cancer cells
much more reliably than humans.
much more reliably than humans
And AI chatbots can behave like patient teachers
And AI chatbots can behave like patient teachers
when we struggle to understand a complex topic,
when we struggle to understand a complex topic
summarising huge volumes of information for us.
summarising huge volumes of information for us
The AI revolution has the potential to enhance and speed up work
The AI revolution has the potential to enhance and speed up work
in many fields, from software programming, to animation,
in many fields from software programming to animation
to law enforcement and journalism.
to law enforcement and journalism
This has pluses and minuses, of course,
This has pluses and minuses of course
but could this extra capacity free us up to do other things?
but could this extra capacity free us up to do other things
Like tackling climate change
Like tackling climate change
or looking after ourselves and each other better?
or looking after ourselves and each other better
As AI advances, governments and regulators will of course
As AI advances governments and regulators will of course
need to make sure it's being used ethically and legally -
need to make sure it's being used ethically and legally
no easy feat.
no easy feat
But will AI take over the world?
But will AI take over the world
Don't forget, AI is a tool,
Don't forget AI is a tool
and even a powerful tool can't take over the world on its own.
and even a powerful tool can't take over the world on its own
It's up to us to decide how we use it -
It's up to us to decide how we use it
or even if we should use it at all.
or even if we should use it at all
Talking speed

You might also like these videos

If you enjoy using our custom video player, why not join thousands of subscribers who have signed up to our free weekly lessons?

By using this website you consent to our cookie policy and website terms and conditions.