In this article (we are still under pressure to call it a blog post) we weigh up the pros, cons and impact of the latest generation of AI generative bots.
This article was written by humans. At the end, you can see a live demonstration, showing how ChatGPT has approached the same subject.
These tools are truly remarkable and some of the hype appears to be justified.
“Some 43% of white-collar workers claim to have used ChatGPT.”
As the fastest growing consumer application in history, now with 100m users (thanks UBS), we ask if ChatGPT and other recently launched generative AI tools are going to have the impact that their promoters are claiming?
Without doubt, tools like ChatGPT, Bing AI and Midjourney are going to massively improve productivity in the white collar sectors that rely on knowledge as their currency.
Science, medical, research, creative and consulting industries will benefit in the way that assembly and manufacturing plants, which made everything from silicon wafers to the Nissan Sunny, did.
In 2011, Erik Brynjolfsson and Andrew McAfee published “Race Against the Machine”. In their analysis, they claimed that the economics of numerous industries would be profoundly impacted by AI-knowledge-based robots. Much the way that claims were made in previous decades, about the impact of automated robots on manufacturing industries.
Japan and Korea (South, obvs) have the highest penetration rates of robots, but have 2.5% and 3% unemployment rates respectively, compared with 6.6% for the Euro area. So the predicted destruction of swathes of jobs in industries that have been automated, have failed to materialise. Yet.
Hold on a sec. Don’t we also keep reading about how productivity is THE big problem in developed economies? The predictions of great jumps in human productivity during the tech booms of the 80s (personal computing), the 90s (productivity software), the 00s (networks) and the 10s (smartphones) have yet to materialise.
Here we go again.
Silicon Valley claiming a new technology will change the world beyond recognition. The media running scared. And academia pushing the technology, to see what the truth is.
And the rest of us are wrestling with how we can use this cool new tech to create value in our businesses and careers.
Remember how Bitcoin, Google Glass, VR, AR and lots of other Rs and Elizabeth Holmes (now in jail), were going to revolutionise our world in profound ways?
What’s it all about, Alfie?
Very simply, generative AI produces text, speech, images and even computer code, based on a simple brief, typed in by a user. First reactions are “wow”. And make no mistake, some of these tools are impressive in the speed and form with which they can create “stuff”.
But after real exploration and application, they are what the plough was to farming, 7,500 years ago. Or steam was to power generation. Or the combustion engine was to transport.
They can make us all go a bit faster. But not always in the right direction.
Milton Friedman, the economics professor, was travelling overseas and passed some roadworks. Looking out of his car window, he was confused. He asked his host “why the labourers were using shovels rather than machinery”. The answer came “it was in order to maintain employment levels”. Friedman’s response?
“Then instead of shovels, why don’t you give them spoons and create even more jobs?”
It’ll huff and puff.
But will it blow any houses down?
Although there is clear nervousness in the culture, media, entertainment and creative industries, many are claiming generative AI is helping to do much of the spade work for workers in these sectors.
The productivity boom. Finally.
But what has tended to happen, as we see from the revolution in manufacturing robots that were installed from the 1970s, is that these technologies should – and a very definite should – allow many of us to do our jobs more effectively.
In the car industry, the main benefits of automation have led to lower defect rates, increased customer safety and longer warranty periods, but not necessarily the cost savings that were predicted. Companies that migrated all of their computer kit to the cloud actually saw costs increase by 15%. It was the speed of access, sharing and aggregation of data that enabled workforces to focus on what they are good at, which in turn delivered the significant benefits of cloud computing.
A war of words.
As mechanical robots have thus far failed to consistently pick up eggs without breaking them, so generative AI tools should be treated with caution.
There are still fundamental flaws in these technologies. So we have to understand their limitations and we mere humans have to learn how to adapt and use them to enhance our work, productivity and the value we add.
Note Moravec’s paradox: “It is comparatively easy to make computers exhibit adult-level performance on intelligence tests or playing checkers,” he wrote, “and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility”.
“A scientific paper was recently published, claiming that chatbots, with their inherent plagiarism, pose a risk to integrity and honesty. It was written by a chatbot.”
Bing Chat has assured users they can safely eat ground glass, claimed the US has had four female presidents and generated false citations. The debacle has confirmed the doubts of that ignored species – the sceptics – who’ve long warned that this model of machine learning is little more than “a statistical word-completion algorithm”.
The written and verbal elements of these AI bots can be like comparing Wikipedia to the Encyclopedia Britannica: superficially, they can help with the simple, quick generation of art, graphics, articles, computer code, melodies and messages. But if you want verified, deep and meaningful content, these bots are not the answer: their algorithms could not for example generate idioms, and certainly not in another language.
But not of numbers.
And it seems that the “T” for the training element of the GPT acronym has not been extended to even basic accounting, let alone marketing ROI forecasting. (Chat GPT stands for “generative pre-training transformer”. Snappy, huh?).
Stuart Cobbe, a chartered accountant, decided to test ChatGPT by entering questions from a sample exam paper on the ICAEW website and then entering its answers back into the online test. ChatGPT scored 42 percent. I for one would not accept my accounts being just 42% correct.
The Google slayer?
What is the difference between Google and ChatGPT? You start off with a problem, request or idea and type it in. Then what happens?
We all know Google will sift through billions of data contained in web pages. It will then present us with a giddying array of results, relying on our own expertise and intelligence to select the one most relevant to what we want. Over time, Google’s own AI works out which data is more likely to be the closest fit to what it has been asked. Google has pioneered the use of AI to search, analyse and generate results for text, image and video searches.
What this new generation of AI tools will do, is aggregate billions of data. Using a pre-trained algorithm (programme) it produces a prediction of the text, image, melody, computer code or animation that it “thinks” matches the user’s brief. The algorithm that has been trained, uses this vast amount of data (and the power consumption of a small village) to predict what you have asked for, based on what has gone before.
And therein lies the key to understanding the limitations of these generative bots.
Even bots need rest.
Imagine Google coming back with this:
Imagine a world without imagination.
“To know is nothing at all; to imagine is everything.”
Anatole France, Nobel laureate for literature
What would have happened if Galileo, Leonardo (da Vinci, not DiCaprio), Magellan, Gutenberg, Aristotle, Newton, Fleming, Brunel, Rowling, Berners-Lee et al had relied solely on what had gone before? Without the power of randomness or creativity, would we have had the inventions of the compass, the suspension bridge, penicillin, Quidditch, the printing press or the world-wide web?
Like the textile Luddites from Manchester, when a new technology which threatens livelihoods emerges, fear spreads. And there is concern among many knowledge-based workers and people in the creative industries about how generative AI might threaten jobs.
However, allowing computers and algorithms to have free rein over what and how we communicate carries some risks. JPMorgan Chase has banned their use over privacy concerns. Oxford and Cambridge University have banned ChatGPT, quoting plagiarism examples.
Get with the programme.
In every client that we work with, we are seeing the possibilities for ChatGPT and its burgeoning rivals.
Adobe Illustrator, Google GA4, Microsoft Office and Amazon Web Services (cloud) have enabled us to create, aggregate, generate and produce all sorts of valuable content for our clients’ brands. This next generation of productivity tools will make us, without doubt, even more productive.
But more importantly, generative AI will allow us to be more imaginative. More creative. Help us to differentiate. Give us more time to think. More time to do what we are good at. More time to add value.
And like all advances in technology, it’s what we do with it that will enable us to be better at what we do.
If we choose to harness it in the right way, how great will it be to free ourselves from the labours of searching, aggregating, typing, selecting, colouring, note taking?
This article was written by humans.
We asked ChatGPT to produce text for us, based on the same brief for the article that we had written. The results are included below.
We will leave you to judge who’s mailing list you would prefer to subscribe to.
Here is a video, showing how ChatGPT produced a similar article to the content above.
This is the text version written by ChatGPT, based on the following brief:
Write me a blog post about the impact of generative AI tools like ChatGPT: we’ve heard it all before: the mass job losses that will be caused by robotics, AI and other technologies. Why have we not seen these predicted reductions in jobs? Has there been significant job displacement caused by AI, robots or other technologies? What will the economic impact of generative AI bots be, how will they impact productivity, jobs and what industries will most benefit or be impacted by generative AI tools like ChatGPT? How should professional and knowledge workers embrace and use tools like ChatGPT?
Chat GPT response:
In recent years, generative AI tools like ChatGPT have become increasingly popular in various industries. While some people have expressed concerns about the impact of these tools on the job market, there has been little evidence to suggest that they will lead to mass job losses. In this blog post, we will explore the impact of generative AI tools like ChatGPT, their economic impact, and how professional and knowledge workers can embrace and use them. One of the reasons why we have not seen the predicted reductions in jobs is that AI and other technologies have created new job opportunities that did not exist before. For example, AI has created jobs in fields like data analysis, software engineering, and machine learning. Additionally, AI has improved productivity in many industries, leading to increased demand for workers in those industries. However, it is also true that some jobs may become automated as a result of AI and other technologies. Jobs that involve repetitive tasks or manual labor are more likely to be automated than jobs that require complex problem-solving or interpersonal skills. While this may result in some job displacement, it is important to note that new job opportunities will also be created as a result of AI and other technologies. The economic impact of generative AI bots is likely to be significant. AI has the potential to increase productivity, reduce costs, and improve efficiency in many industries. For example, in the healthcare industry, AI can be used to analyze medical images, identify diseases, and develop treatment plans. In the financial industry, AI can be used to detect fraud, improve risk management, and automate routine tasks. The impact of generative AI tools like ChatGPT on productivity and jobs will vary by industry. In industries where there is a high demand for creative problem-solving and interpersonal skills, such as marketing, design, and education, ChatGPT can be used to enhance the work of professionals in those fields. For example, ChatGPT can be used to generate marketing copy, design concepts, and educational materials. In industries where there is a high demand for manual labor or routine tasks, such as manufacturing or transportation, ChatGPT may be used to automate certain tasks, leading to some job displacement. Professional and knowledge workers can embrace and use tools like ChatGPT by using them to enhance their work, rather than replace it. ChatGPT can be used to generate ideas, provide insights, and streamline processes. For example, a marketer can use ChatGPT to generate ideas for a new campaign or a teacher can use ChatGPT to develop lesson plans. In conclusion, while there may be some job displacement as a result of generative AI tools like ChatGPT, the overall impact on the job market is likely to be positive. AI has the potential to create new job opportunities and enhance productivity in many industries. Professional and knowledge workers can embrace and use tools like ChatGPT to enhance their work and stay competitive in the job market. It is important to recognize the potential of AI and other technologies to improve our lives and the economy, while also addressing the challenges and concerns they may pose.