Let's chat
© 2024 CX Lavender
PoliciesMythbusting AI
It's the buzzword on marketers lips – but does AI really have what it takes to make us all obsolete?
This week we probe into the truth, morality and inner workings of AI. Can and will it really take our jobs?
Ben Daley
Strategy
Trending
6 minute read
Advances in technology are lightning fast, yet somehow glacial at the same time. It seems like no matter how much we invest in imagining what the future holds, it never fails to fall short of expectations.
Artificial Intelligence is no exception.1
On the one hand, it’s the next digital frontier, promising nothing short of an evolution of humanity2 as we increase our interconnection with the devices that surround us. Intelligent machines complement humanity, enriching our lives with thoughtful interactions and pre-empting our needs at every turn.
On the other hand, it’s yelling, 'Hey Siri, skip this song!' because your phone is blaring that 2014 U2 album and is convinced you’re a big fan.
But as much as the promise of AI has not yet eventuated, enough buzz has been generated and enough products have entered our everyday lives that we have to ask: what is reality and what is still just sci-fi?
Artificial Intelligence is no exception.1
On the one hand, it’s the next digital frontier, promising nothing short of an evolution of humanity2 as we increase our interconnection with the devices that surround us. Intelligent machines complement humanity, enriching our lives with thoughtful interactions and pre-empting our needs at every turn.
On the other hand, it’s yelling, 'Hey Siri, skip this song!' because your phone is blaring that 2014 U2 album and is convinced you’re a big fan.
But as much as the promise of AI has not yet eventuated, enough buzz has been generated and enough products have entered our everyday lives that we have to ask: what is reality and what is still just sci-fi?
In short
- Mislabelling all smart tech as ‘AI’ creates confusion about what AI is, and how far it’s come.
- Hype has set unrealistic expectations for what AI can do now, or will be capable of doing in the future.
- It is unlikely that our jobs will be replaced by armies of androids any time soon.
Three truths about AI
#1: AI is not a ‘thing’
AI is used as a blanket term to cover all things ‘smart computer’. From the processing of big data to the voice activation of a smart home device, people are amazed at the many applications of AI in our lives.
But this is a misuse of the term, and helps confuse progress in artificial intelligence with advances in computing. Here are some helpful definitions to help distinguish it:
Artificial intelligence3 is specific to the quest to mimic human intelligence in a machine – a computer that can think the way we think, but with the power of Intel Inside™. We are a long way off having conversations with Data from Star Trek – any chatbot or voice command system currently needs to be trained using data, and is more of an application of machine learning than true AI.
Machine learning is the processing of large quantities of data to allow a sophisticated computer program to increase the probability of a desired outcome or identify the probability of any one outcome. The first is applied in self-driving cars as they improve road safety. The second helps you decide if it’s a good idea to bring an umbrella.
Automation connects inputs and outputs in a way that streamlines work. Inputs might require action – 'OK, Google, play Rocketman' – or might be triggered by a preset condition – when the alarm goes off, turn on the bedroom lights. A lot of the low-hanging fruit called 'productivity' stems from automating mundane tasks so that the real asset of a business – human brainpower – can be freed up to work more creatively on problem-solving.
But this is a misuse of the term, and helps confuse progress in artificial intelligence with advances in computing. Here are some helpful definitions to help distinguish it:
Artificial intelligence3 is specific to the quest to mimic human intelligence in a machine – a computer that can think the way we think, but with the power of Intel Inside™. We are a long way off having conversations with Data from Star Trek – any chatbot or voice command system currently needs to be trained using data, and is more of an application of machine learning than true AI.
Machine learning is the processing of large quantities of data to allow a sophisticated computer program to increase the probability of a desired outcome or identify the probability of any one outcome. The first is applied in self-driving cars as they improve road safety. The second helps you decide if it’s a good idea to bring an umbrella.
Automation connects inputs and outputs in a way that streamlines work. Inputs might require action – 'OK, Google, play Rocketman' – or might be triggered by a preset condition – when the alarm goes off, turn on the bedroom lights. A lot of the low-hanging fruit called 'productivity' stems from automating mundane tasks so that the real asset of a business – human brainpower – can be freed up to work more creatively on problem-solving.
#2: The dream of AI has been well sold
Perhaps it’s cynical for marketers to criticise the work that AI evangelists have done spruiking their cause. After all, we spend our days identifying the needs of people and then finding novel ways to connect our products to these needs, sometimes with very liberal ‘perfect world’ scenarios.
Were we to turn our marketing heads to the task of selling AI, however, our approaches would almost undoubtedly intersect:
Cynical or not, we can’t begrudge the approach for being effective. Business leaders have bought in en masse, looking for ways to unlock additional output from tech investments, and the noise from the hype machine has made it hard to pick what’s real, realistic, or really far off.
Were we to turn our marketing heads to the task of selling AI, however, our approaches would almost undoubtedly intersect:
- Lead with the promise of the future state of AI – the glorious utopia of the future, built on the backs of robotic muscle as humans guide government and enterprise to solve real problems.4
- Introduce the fear of falling behind – if you’re not investing in AI solutions yesterday, your business cannot possibly survive and thrive in the future. ('How much should you be investing? See this handy pie chart that visualises a selection of hand-picked budgets for emphasis!').
- Provide a concrete action that can be taken today – tools and products that you can subscribe to right now to better manage your calendars, saving your employees valuable time. Or better yet, a future-focused approach, built on ensuring you capture every single current and potential interaction with your customers as data points, in case they provide the key in training your next bespoke machine-learning solution, one day.
Cynical or not, we can’t begrudge the approach for being effective. Business leaders have bought in en masse, looking for ways to unlock additional output from tech investments, and the noise from the hype machine has made it hard to pick what’s real, realistic, or really far off.
“The noise from the hype machine has made it hard to pick what’s real, realistic, or really far off.”
#3: Robots are (probably) not coming for your job
Some years ago, I sat with a colleague in a presentation of the power of IBM’s Watson to automate and manage marketing campaigns. We were shown the computer searching out audiences, refining and testing headlines, changing creative based on weather, and optimising the media buys based on live data and purchase patterns.
We went back to the office to polish our resumes, waiting to see what our new machine overlords’ position on the four-day working week was. But so far, we have yet to see this dystopic* vision of the future of work unfold.
*The robot revolution, not the four-day week.
We went back to the office to polish our resumes, waiting to see what our new machine overlords’ position on the four-day working week was. But so far, we have yet to see this dystopic* vision of the future of work unfold.
*The robot revolution, not the four-day week.
In summary, here are some answers to questions you may have:
Have machines replaced jobs? Yes.
Will intelligent machines continue to replace jobs? Certainly.
Will artificial intelligence lead to mass unemployment as only the programmers survive? Highly unlikely5. While many repetitive and mundane tasks can be automated, knowing when and how to run these tasks still requires a lot of human insight and oversight. For instance, many checkout attendants have been replaced by self-service terminals, but how many customers actually like it? If there’s a manned supermarket checkout available and I’ve got a full trolley, I always opt for assistance.
Could a cleaner be replaced by a robot vacuum? Would the robot vacuum be able to distinguish between a lost diamond earring that fell on the floor and dust bunnies?
Could a teacher be replaced by education software? Would that software be able to identify a student who was struggling with a concept and help them understand? Or to help a particularly gifted child explore ideas that the curriculum says they’re not ready for?
Could a marketer be replaced by Watson? For a company with sufficiently deep pockets… probably not. No matter Watson’s ability to run optimisation scripts when triggered by specified events, it can’t come up with a truly moving creative idea, understand how people might use or engage with a product, nor craft a meaningful headline. Remember, it works on existing patterns to ‘think’, so it can’t create truly original ideas. It can’t take inspiration from looking out the window on its bus ride to the office, or having spit-ball conversations with mates.
Neither believer nor skeptic
It might, at the conclusion of this short list, feel as though I am in some way anti-AI, or disbelieving of the possibilities that AI could unlock in our future.
This couldn’t be further from the truth.
I happily use automation in my daily life, and look forward to a time in the future where we’re able to achieve so much more by focusing our minds on the things that make us uniquely human, rather than some fleshy computer. Emotions, humour, optimism, hopefulness… AI can’t replicate – let alone replace – any of this.
But as much as the future is still a blank page, I don’t believe the opening paragraphs will introduce artificial intelligence or machines as the protagonist any time soon.
Until such a time, take any promise or warning of the power of AI with a grain of salt.
The good, the bad, and the ugly of real-world AI
Running the gamut from making our lives easier to making the world worse means that AI has a lot to answer for. Let’s have a look at the best and worst of AI experiments.
The good: Face ID
Face ID is AI at its best, allowing us to go about our day a little more effortlessly. Apple says you’ve got a one-in-a-million chance at tricking Face ID. While reality might place a few roadblocks in its way (think face masks), we still count Face ID as an overall win for AI.
The bad: ball or bald head?
When Inverness Caledonian Thistle FC moved to livestreaming their games, they decided to use an AI-powered camera system to track the ball, eliminating the need for a camera operator. But the system ultimately failed when it couldn’t differentiate between a soccer ball and the bald head of a linesman. Oops.
The ugly: Tay’s day on Twitter
Microsoft’s chat bot experiment turned ugly in less than a day. The Twitter bot named Tay started by replying to users with excited tweets but soon took a sharp turn into racism and anti-feminism. The most offensive messages, sourced from the nastiest Twitter had to offer, were subsequently deleted and Tay was shut down by Microsoft – only 16 hours after its launch.
The good: Face ID
Face ID is AI at its best, allowing us to go about our day a little more effortlessly. Apple says you’ve got a one-in-a-million chance at tricking Face ID. While reality might place a few roadblocks in its way (think face masks), we still count Face ID as an overall win for AI.
The bad: ball or bald head?
When Inverness Caledonian Thistle FC moved to livestreaming their games, they decided to use an AI-powered camera system to track the ball, eliminating the need for a camera operator. But the system ultimately failed when it couldn’t differentiate between a soccer ball and the bald head of a linesman. Oops.
The ugly: Tay’s day on Twitter
Microsoft’s chat bot experiment turned ugly in less than a day. The Twitter bot named Tay started by replying to users with excited tweets but soon took a sharp turn into racism and anti-feminism. The most offensive messages, sourced from the nastiest Twitter had to offer, were subsequently deleted and Tay was shut down by Microsoft – only 16 hours after its launch.
on AI’s moral compass
Here’s another reason AI isn’t coming for your advertising job: its struggle to resolve an ethical dilemma. Take the time a model predicted 'criminality' and generated racist results. How could AI compete with human minds when creating a great ad isn’t just about the right idea, but executing it within ethical boundaries?
References
- John Horgan, Will artificial intelligence ever live up to its hype? (4 December 2020) Scientific American.
- Janna Anderson, Lee Rainie, Artifiical intelligence and the future of humans (10 December 2018) Pew Research Centre.
- Microsoft Azure, Artificial intelligence (AI) vs machine learning (ML) (Date unknown) Microsoft.
- Jason Mars, The antidote to the hype, noise, and spin of artificial intelligence (25 March 2021) Forbes.
- David de Cremer, Garry Kasparov, AI should augment human intelligence, not replace it (18 March 2021) Harvard Business Review.
