First year of university.
A mid-morning Accounting for Engineers tutorial in a lab full of Apple computers, each of which cost more than my car.
The lanky tutor sported a waist-length ponytail. A Thai-dyed shirt, suggesting questionable judgement or colour blindness, purchased for peanuts from a roadside vendor somewhere in lesser-travelled South-East Asia. And a truly spectacular hangover.
The computer lab was not his native environment. A hastily improvised emergency measure, creating a provision for distraction to avoid the very real prospect of his breakfast being contra’ed out via a technicolour yawn at the merest mention of a debit or credit.
Attendance was meagre. Most of the class had been at the same rowdy after-party as the tutor the previous evening. An anarchist ska-punk band had raised the roof of the university bar, performing a fast-paced set for the writhing mob of horny sweaty students in the mosh pit.
Inspired by the lead singer’s throwaway comment that he could drink “a truckload of beer right about now”, during the encore one of the more ethically flexible audience members had slipped away. He returned shortly afterwards with a delivery truck liberated from the loading dock of a nearby alcohol superstore.
An avalanche of humanity streamed out of the bar and onto the neighbouring playing field at the end of the gig. With impeccable showmanship, the student anarchist threw open the rear door of the truck to a huge chorus of hero worship and adoration.
For one tragicomic moment, there was stunned silence as the cargo was revealed to be endless pallets of rainbow-coloured alcopops. Then as one, the crowd collectively shrugged, figured any free drink was better than no drink, and cheered once more.
The lolly water libations finally ran out as the sun peeked over the horizon. Revealing an early morning mass of bodies scattered across the playing field amidst a sea of empty pre-mixed bourbon, rum, and vodka-based beverage bottles.
A few hardy souls made it to the tutorial. Wearing yesterday’s clothes and last night’s hangover.
They were in store for an unexpected treat.
The tutor showed the students how to run a beta version of a computer program called Netscape and use it to open something called Yahoo! It turned out Yahoo! was a website, one of roughly 20,000 that it proudly claimed to have indexed across something inexplicably named the World Wide Web.
I remember experiencing a feeling of initial puzzlement, followed by stunned amazement: this will change everything!
As I clicked from site to site, I realised that what I was seeing granted anyone the ability to instantly find the answers to just about any question. It was as if I could ask unlimited questions to a smart person who had already memorised the entire content of the local library.
No more needing to borrow books or reserve academic journals, competing with my classmates for access to the scarce resource materials required for our studies. Come to think of it, probably no further need for libraries at all.
No more having to sell a kidney or sacrifice a firstborn child to pay for the overpriced academic textbooks prescribed for each course of study. Come to think of it, probably no further need for bookshops at all.
Seemingly In the blink of an eye, I found myself seated alone in the computer lab. The sun had gone down, and I had lost the best part of a day down what would later become known as the internet rabbit hole. Missing lunch, and then dinner, as I traversed hyperlink after hyperlink.
I distinctly remember the feeling of utter certainty I possessed while driving home that night, that this was a genie that could not be put back in the bottle. A game changer likely to rank alongside the invention of fire, the wheel, the printing press, and the telephone in terms of impact.
In the 20+ years since then, I never again experienced that feeling of having glimpsed the future. Until now.
There have been a few false starts. Ideas that initially sounded promising, but further investigation found them wanting as more of the same, or a technological dead end.
Outsourcing, swiftly followed by offshoring. Resulting in the emergence of a global marketplace for skilled labour, and the inevitable race to the bottom in the market prices those skills can command.
Sequencing the human genome. A grand achievement in its own right, but the overhyped promises of ending disease and solving mortality have so far failed to materialise. A stark reminder that data only becomes useful information with the application of knowledge and understanding.
Clustered computing. Grid computing. Cloud computing. Granting the ability to throw ever more hardware at a problem, breaking big jobs down into multiple smaller jobs performed in parallel, using computers maintained by someone else.
Distributed ledgers. Solving the same old problems with shiny new technology, usually not as well. The genuine use cases that could be solved only by such an approach are few and far between.
Quantum computing. A fascinating idea, that we don’t yet possess the technology to explore.
What was sufficiently interesting that I experienced that same combination of intellectual excitement and existentialist dread? The public trial of OpenAI’s ChatGPT robot.
The premise is that using natural language, you ask a robot to write something for you. The robot then attempts to create something that a meets your need.
Essays.
Reviews.
Songs.
Stories.
Even computer programs.
Which sounds mildly interesting, until you start to play with it.
I asked the robot to write me a limerick about personal finance. A second or two later, it responded with:
There once was a man named Dave
Who loved to spend all he had save
But one day he learned
That money must be earned
Now he’s saving and investing each day.
Next I changed my request for a personal finance limerick to a haiku. The robot swiftly returned with:
Money flows like a river
Savings grow with careful planning
Wealth blooms like a tree
Changing the request again, I asked for a fable instead of a haiku. The tireless robot penned this:
Once upon a time in a small village, there lived a wise old fox named Finn. Finn was known throughout the village for his wisdom and his knowledge of personal finance.
One day, a young rabbit named Robin came to Finn for advice. Robin had just received his first paycheck and was not sure how to manage his money.
Finn sat Robin down and explained that it was important to save some of his money for the future. He advised Robin to create a budget and stick to it, and to avoid unnecessary spending.
Robin listened intently and thanked Finn for his advice. He went home and created a budget, and began saving a portion of his income each month.
As the months passed, Robin saw his savings grow and was able to make important purchases, such as a new home and a reliable car.
One day, a great storm hit the village and many homes were destroyed. Robin was grateful that he had saved his money and was able to rebuild his home without any financial stress.
The other animals in the village were impressed by Robin’s wise financial decisions and began seeking advice from Finn as well.
From that day on, the village was known for its financially savvy inhabitants, all thanks to the wise old fox and his advice on personal finance.
Deciding that I had been too easy on the robot, I was determined to defeat it. I asked it to write a personal finance song in the style of Michael Jackson. It instantly output:
Verse 1:
Money, money, money, oh how it flows
Gotta keep track, gotta watch how it goes
Bills to pay, savings to grow
Personal finance, gotta keep it in control
Chorus:
Don’t stop ’til you budget, don’t stop ’til you save
Gotta plan for the future, gotta be financially brave
Don’t stop ’til you invest, don’t stop ’til you thrive
Personal finance, keep it alive
Verse 2:
Credit score, credit cards, oh what a game
Gotta use ’em right, don’t let ’em go to waste
Debt to pay off, investments to make
Personal finance, gotta keep it awake
Chorus:
Don’t stop ’til you budget, don’t stop ’til you save
Gotta plan for the future, gotta be financially brave
Don’t stop ’til you invest, don’t stop ’til you thrive
Personal finance, keep it alive
Bridge:
Gotta stay on track, gotta be disciplined
Gotta be smart with your money, gotta be minimally financed
Gotta keep on saving, gotta keep on growing
Gotta keep on dancing, personal finance, it’s all worth knowing
Chorus:
Don’t stop ’til you budget, don’t stop ’til you save
Gotta plan for the future, gotta be financially brave
Don’t stop ’til you invest, don’t stop ’til you thrive
Personal finance, keep it alive.
Next, I asked the robot to write an upbeat 500 word blog post detailing the secrets to personal finance success. The response read like 95% of the personal finance content that appears on Sovereign Quest each week. Swapping blog post for tweet in the request yielded the following:
- Start by setting clear financial goals and creating a budget to track your spending.
- Cut unnecessary expenses and prioritize savings.
- Invest in a diverse portfolio to maximize your returns.
- Educate yourself on financial topics and seek professional advice when needed.
- Stay disciplined and consistent in your money management habits.
#personalfinance #success
At that point, I sat back in amazement.
Instead of playing with the robot like a novel toy on Christmas morning, I attempted to put it to productive use.
I asked it to write a strategy for a specific niche industry problem one of my teams at work was attempting to solve. The response was not perfect, but it was a credible attempt. To be perfectly honest, it was better than the first couple of drafts a team of experienced and expensive management consultants had produced.
I tried a couple more real-world business applications. The output was about the same as I receive today from freelance technical writers based in low-cost locales like Kuala Lumpur or Lahore. I am guilty of timesaving by outsourcing the creation of bullshit business documents and low-value presentations, whenever I suspect nobody will read or maintain them. This technology may save me a little bit of money, while signalling the beginning of the end for one of the few genuinely location-independent jobs.
As I thought about the potential applications of the technology, I realised it had potentially vast implications.
Cheating on academic assessments or work assignments will become even easier. Already my elder son tells stories of cheating becoming the default in high school, where one kid distracts the teacher as another WhatsApps a hastily taken Microsoft Lens image of the exam paper to their peers in other classes. The members of the subsequent classes then know exactly what to study, gaming a flawed education system focused on teaching standardised exams rather than understanding course content.
Customer service, without the comprehension difficulties that plague the current generation of chatbots. The robot was able to comprehend almost all the questions I asked it, even when my phrasing was clumsy or contained spelling mistakes. Changing the wording often changed the answer, so there is a way to go yet, but the direction of travel was clear.
Then I thought about how much of the white-collar working world involved producing and summarising low-value reports or repetitive analyses. The kind of unoriginal thinking, paint-by-numbers procedural busywork, that constitutes much of what consumes many hotdesk occupant’s working week.
Before I come off sounding like a love-struck fanboi, it is important to recognise the limitations of the technology. The robot produces a lot of plausible-sounding nonsense.
It isn’t capable of thought, just regurgitating what was contained in its training data. Confusing causation with correlation. Conflating concepts. Lacking knowledge, nuance, and personality. In many cases, it resembles an armchair expert who possesses just enough knowledge to be dangerous.
Using personal finance as a use case, the robot personifies the endless echo chamber of recycled ideas which have been pilfered and plagiarised so often that they have become conventional wisdom.
The promise of what this technology will offer in the future in equal part excites and terrifies me. Much like the early internet I encountered during that hungover tutorial, that future promise far exceeds the realities of the current implementation.
Much like that early internet, I can already start to see just how transformative it has the potential to become. The white-collar world has long been a safe harbour for well-remunerated workers to finance a comfortable lifestyle endlessly moving data, producing slide decks, torturing spreadsheets, and writing code.
Those workers are about to experience first-hand what their agrarian, mining, and production line working forebears felt like a generation or three ago. It will be fascinating to watch the evolution.
References
- Microsoft Lens (2022), ‘Microsoft Lens: PDF Scanner‘
- OpenAI (2022), ‘ChatGPT‘
- Yahoo! (1996), ‘Yahoo! Search engine‘ [via Wayback Machine]
Featured by
The young investor 5 December 2022
Thank you for sharing this, I have been talking with the bot for the last few hours and this is definitely going to replace Google (and my job as business consultant soon after that). Never been so impressed by any other tech before.
{in·deed·a·bly} 5 December 2022 — Post author
Thanks young investor.
I think technologies like this will extend rather than replace Google, at least in the near term.
The business consulting thing is an interesting one.
The robot would be able to tell you how everyone else does it, based upon training data (github, stackoverflow, quora, allrecipes, gartner, etc… basically the collective wisdom of the public internet). However, that training data will contain all the shortcomings of the internet… the buggy code on github, the unsubstantiated claims of wikipedia, the opinionated hellscape of reddit comments or social media, etc. So in theory it might be faster and cheaper, but not necessarily better, than hiring a consultant who possesses broad industry knowledge.
What it won’t be able to do (initially at least) is pander to the fragile egos of the C-suite, which make every firm believe they are unique and special. Indeed, consultants have a vested interest in prolonging those behaviours. In all but a few very core competencies, most firms are very alike, so commodity backoffice services are potentially vulnerable.
But where genuine unique thinking occurs, there won’t be training data available for the robot to draw upon. That said, most innovation is incremental enhancement rather than revolutionary game changing. I would expect in time the robots will develop the ability to iteratively join dots and extrapolate, which is at a fundamental level what machine learning is all about. The song lyric example above illustrates this at work.
Which is basically what humans do: stand on the shoulders of giants, tweak things a little bit to go faster or work better, and repeat.
Those genuine original thinkers will be ok. In truth that is a skill rare enough to be considered a superpower, seldom observed in the wild.
The rest of us, who ceaselessly reinvent the wheel in code, spreadsheets, or professional business voice writing will be observing dark clouds gathering on the horizon, and perhaps experiencing a feeling of impending doom.
Leading us to ask ourselves honestly about the value proposition our collective skills and experience bring, and why an employer would need to continue paying us to perform our services when capabilities such as this are available for free. If the best we can come up with is a question of good versus good enough, then that forecasts an uncomfortable road ahead.
The young investor 5 December 2022
Thank you for your answer, it is true that this bot can’t please the ego of my boss (my primary role) for now. Btw, I’ve fed your entire answer to it and apparently you are on the same page.
Tomas 11 December 2022
Wow! This technology is a true game changer.
Retire by 50 – Guns’n Roses style
Argue for and against retiring at the age of 50 by writing a song in the style of Guns’n Roses.
A 5 December 2022
Always so thought provoking, indeedably. Your blog is an oasis of honest quality. Thanks for writing it.
My own Q&A session with Open Ai has mostly revolved around AI investment ideas, alas it seems offering up ‘the next big thing’ isn’t an easy task and the investment industry’s propositions of funds consisting of obvious blue chips with fancy marketing brochures and equally fancy management fees doesn’t fill me with confidence.
‘The Bot’ suggests Clarifai, Cogitai and Bonsai for start ups and MSFT, IBM and GOOG for blue chips.
The answer to my more existential personal finance questions were unsettlingly vague
“Will my investments be worth anything when you take over the world?”
Yes, your investments will continue to be worth something. However, we cannot guarantee that the value of your investments will remain the same.
{in·deed·a·bly} 5 December 2022 — Post author
Thanks for the kind words A, glad you’re finding things to your liking.
“The next big thing” is an interesting challenge. If you think about it, the robot’s responses are drawn from its training data, so asking it to predict the future is likely to end in disappointment. This could potentially play out in several ways:
1. It regurgitates sponsored propaganda about the next big thing, in the same mindless way that the mainstream media and financial press present press releases as fact or run flattering advertorials rather than reasoned evaluations. This is the likely outcome, a continuation of the high-fee high-churn fashion chasing active investment approach favoured by those who make a living managing wealth out of the clients’ wallet into their own.
2. Its model is evidence based and predictive. Extrapolating trends. Identifying repeating patterns. Using statistical methods to take an educated guess on what happens next. The sort of thing quants have been attempting to do for a couple of decades, only faster and cheaper. We see those types of algorithms at work, with mixed results, in fraud detection systems today.
3. It identifies the contrarian play. Figuring out the momentum of the herd, and then making money by selling into them. The same approach long adopted by the large investment banks, where the research analysts steer the investing public in one direction, while porous “Chinese” walls lead to the bank’s trading desks magically taking positions to take advantage of that direction of travel. We already see some examples of this, for example free trading platforms who front run their customers orders, or high frequency trading platforms placing and cancelling phantom orders to move the market advantageously.
4. It is used to create noise and influence market behaviours. Planting industrial volumes of AI generated fake tweets or reddit posts on /r/WallStreetBets for example, to generate noise and move markets. Similar to what we observe with Amazon, Deliveroo, TripAdvisor, or Yelp reviews, only on a grander scale.
There is more, but markets are more about behavioural psychology than anything else. My tip is the robots don’t have to pick winners, they can manufacture them. Picture a denial of service botnet army, but scale it out to AI robots capable of creative writing and instantly generating content at scale. It’ll happen faster and more often than any human moderator or content curator can combat.
Ben Swain 11 December 2022
Great discussion as ever. I read something on this topic recently which gives a similar view in many ways. ChatGPT needing to be treated with caution, as a tool not a silver bullet as it is essentially regurgitating etc.
https://stratechery.com/2022/ai-homework/
What I like though is the pitch that this could ultimately be a watershed to teaching and training robust critical thinking in the digital age. Sorting the wheat from the fake news.
{in·deed·a·bly} 11 December 2022 — Post author
Thanks Ben Swain.
I think where this goes is a similar arc to production.
Back in the day making things was manual, slow, error prone, and labour intensive.
Then technology followed by automation sped things up, reduced unit costs, and improved quality by removing much of the unpredictable human error element. Those who did things manually became technically redundant and largely died out as a viable profession, the few remaining outliers becoming a niche artisan throwback, rare enough to be a rich person’s novelty or a tourist attraction. The new skill was designing/managing the production process, rather than doing the production.
Now we’re approaching a point where that designing and managing will itself become technically redundant. We’ll tell a robot what we want, and it’ll do it for us. Not quite yet at the Star Trek replicator end of the spectrum, but that is the direction of travel. Perhaps not yet for physical goods, but the written word or computer algorithm it is a certainty.
This generation of robots can write us a strategy, but they can’t tell us what goals the strategy seeks to accomplish, nor the success criteria via which we will recognise if we have attained them. Both are subjective, and require original thought… the first couple of times they are asked for. But once the innovators have had their say, the training data exists for the followers to blindly line up and ask the robots to copy the answers.
Ask yourself how many genuinely original ideas are there?
In fiction, there are maybe a half dozen accepted types of story, each of which has a formulaic plot line. It is one of the reasons why we enjoy it, finding comfort in the familiarity.
In personal finance, there are less than a dozen distinct ideas.
In computing, there are just a handful of basic operators involving decision statements and control of flow. All the rest is combining those building blocks into algorithms, which hard won experience see evolve into infinitely replicated design patterns and best practices.
In financial planning, the general timescales and direction of travel are the same for nearly all of us. The method for succeeding involves investing, compounding, and time for everyone. The tailoring of a plan to a person is mostly tinkering around the edges, making us feel heard while providing therapy to get us to follow proven good habits.
So the newly emerging skill becomes one of being able to articulate the ask for the robot, instead of strategising and coding and testing and implementing.
It’ll be faster.
It’ll be cheaper.
It’ll improve quality.
It’ll involve far fewer people in the white collar world, where if we’re honest with ourselves, many of us spend much of our time performing repetitive low-value tasks that pander to egos and perpetuate corporate culture/politics. Producing documents nobody will read. Summarising those documents into slide decks for forgetful inattentive executives. Planning meetings before and status updates after. Rinse and repeat.
Not so much a brave new world as the natural evolutionary continuation of the old one. Which reflects the irony in people flapping about the societal cost of the aging population and demographic time bomb, yet nobody commenting on the fact that at the macro level few of those exiting the workforce are missed for long (if at all).