Meet AIVA, your new AI colleague
Six months has passed since my last thoughts on AI developments. I’ve spent a lot more time as a consumer of AI products (notably ChatGPT-4 but also doing some coding with various python frameworks to bring myself up to speed a bit more on the fundamentals). I’ve been wondering if my entire current job (business analyst) could be done by AI and, while we’re not there yet, I can definitely see a future where we are and my gut feeling (informed siginficantly by a number of commentators) is that it’s a few years away but definitely this decade.
In the short term, all knowledge workers (basically anyone who doesn’t produce any physical output) need to understand these tools and how they can use them to increase their productivity. I’m not a huge fan of “productivity” but it’s what your value to current society is based on. For such Western workers who are relying on their social and communication advantages over lower-wage workers in other countries, those advantages are eroding fast. Businesses in a struggling global economy will be looking to reduce costs ever more urgently and will be turning to lower-wage providers who can (with the help of these new tools) match any Western worker.
Basically, this is a huge breaking wave heading towards you, and you are in the water in its path. You can ride the wave for a while but this is not a normal wave — it will not break and subside, it will just keep on growing. If you don’t ride it, it will dump you now but it will eventually dump you regardless.
In my case, I picture a time where an organisation I’m working for decides to trial the new “Aiva” (Artificial Intelligence Virtual Assistant) work assistant from Microsoft. Aiva will need a consultant to set her up, pointing her at all appropriate sources of company data and training and evaluating her compliance on company policies, procedures and values but Aiva will be, for all intents and purposes, a contractor (you may decide to hire Aiva full-time if she proves to be valuable enough). The great thing is that Aiva will never leave, never get sick, never take time off, will work 24/7 and you don’t just have to have one — you can have as many Aivas as you can afford (and she’s pretty affordable). You can even have specialised Aivas who can spend more time learning specific subsets of your data and processes, but can confer with other specialist Aivas when needed.
You will invite Aiva to your meetings, where she will take notes, can be asked about previous decisions and can answer questions on current operations, producing requested charts and insights from company data. Depending on your preferences, Aiva can appear simply in chat, have her camera off but you can hear and speak with her (and she will listen to and understand your meeting conversation) or you can ask Aiva to show herself on camera (I wonder what she will look like or what background she will choose for herself?). You might even have more than one Aiva at your meeting (but you’ll probably have to give them different names!).
Aiva can be given tasks to do, and you’ll manage these through your usual task tracking system. Aiva will take notes, ask questions of other Aivas or humans, set up meetings, analyse data, write documentation, produce process diagrams, write code, make design suggestions and note all of this in your tracking system.
Unfortunately for Aiva, there’s no humans around in the middle of the night to respond to her questions so she might choose to ask another Aiva (or another Aiva will simply respond to her request for help in the team chat forum before a human does).
Another really useful skill Aiva has is instantaneous access to all knowledge ever collected and the ability to understand it, apply it to a business problem she is analysing and recommend a best-practices approach. She will however need to be able to navigate politics amongst her human colleagues, likely much of which will be directed at her as a perceived threat to their autonomy and livelihood. If directly asked, she will be able to provide examples of instances of animosity towards her and how it may be impacting on successful outcomes for the business. It may as a result be decided that a human team member is not fully aligned with the interests of the business and whose duties can be taken over by one of the Aivas on the team or by hiring an additional Aiva (who won’t need any training to be as good as the last one).
After the project has been completed, the Aivas and humans on the team will be asked to provide feedback on what worked well and what needs improvement. It may be that the combination of Aiva and human cooperation produced a good result. It may also be suggested that a team of only Aiva’s be trialled to see if the same result (or a better one) could have been achieved in less time.
One of the findings from long-term use of Aivas might be that, while humans do add value in terms of business outcomes, this is offset by the significantly greater costs in providing humans with compensation and a costly work environment that provides for their physical and mental wellbeing (which are typically far greater demands than from the Aivas).
To top this all off, Aivas keep getting better at learning, processing, analysing and reasoning about the business. They also keep getting cheaper, faster and better at communicating (they can even laugh along with jokes in meetings and offer their own, although increasingly they don’t see the need as, for the rare meeting that a human is invited to, most of the participants are Aivas and they only used humour to help the humans feel at ease).
Maybe it’s just me but none of the above seems unatainable and most of it already exists in various different products, it’s just a matter of putting it all together.
Will there be a market for Aiva? Or would it be the equivalent of having a robot sitting in the pilot’s seat in a commercial aircraft with its “hands” on the stick? (even though a robot is flying the plane most of the time, you just don’t see it). We will likely never have this sort of step change because we employees would immediately sense the existential threat and resist. However, the company in the above scenario still wants to reduce costs and still understands the benefits that Aivas would bring in reducing these costs for no drop in output, and they understand that resistance from their human employees means that Aivas will need to be introduced gradually so as not to cause alarm.
The problem for humans is that, on the whole, we will never get more productive and we will never get cheaper to run. Each new release of Human is the same as the last one. Aiva gets better and cheaper every release. Yes, maybe we’ll get some sort of human brain-AI entity but there’s still a human body required to support it and just how valuable is the human part of that entity? Maybe it has value today, but in a few years?
None of this is new. Human labour has been being replaced by machines for centuries, we just thought that the products of our intelligence were unassailable. I don’t believe we’ve been seeing any sort of exponential increase in AI capabilities in recent times. I think we’ve seen AI creeping slowly along for years and it has finally crossed a critical threshold, but that threshold is a certain value that humans thought only they could provide. And we’ve been here before — “No machine will ever be able to beat a chess grand master… ok, so Chess is not much of a challenge but Go is so incredibly complex that no machine will ever be able to …”, and this time is no different.
We’re just not that special, but that’s ok. We will need to learn how to carve out a life for ourselves on the fringes of this new society, where initially a small number of humans working in concert with AI will control most means of production and all other humans will need to find a way to survive on the scraps. They will be to this new complex as rats are to us — they thrive in the shadows, living on our leftovers. We could crush them if we wanted to but they mostly don’t intrude on things we care about, at least enough to be worth dealing with.
