Goldman Sachs, those saints of propriety and discretion, and many others are now proclaiming the end of many professions due to AI. The emperors are now admiring their new clothes in the mirrors and trying to make a buck out of it.
Goldman Sachs are now predicting the loss of 300 million jobs. Euronews has a long list of professions exposed to AI impacts. …And so on. You’d think this was all going to happen yesterday.
It didn’t. Pity the clickbait superficiality can’t live up to the facts. Let’s start with the very basics of current tech in a very basic context with AI.
These are the standard arguments for AI vs jobs:
- AI can write. Sure, it can, and it can write competently. That tech was happening separately years ago and was simply grafted onto AI. The AI learned how to do what other tech was already doing. Whether or not it can write well enough to be read in a semi-literate culture is something else.
- AI can do graphic arts and fake imagery. So it can. So can much older technology. Deepfakes predate current AI by about a decade. Guess where AI learned how to do fakes. It can manage objects, but can it do composition a la Renoir? Unlikely at best.
- AI can do maths. So can an abacus. Can it identify the mathematical problems and information needs, not just do the answers from acquired metrics? …Because that’s what just about every other form of computational tech does. You don’t need AI, let alone chatbots, to do that.
- AI can do news analysis. Well, can it? Humans seem pretty bad at news analysis, and AI would have to learn from somewhere. In this case, AI would have to learn from mainstream media how to analyze news. How “threatened” can you feel?
As you can see, AI is really just doing things done by other technologies and applying their systems. AI’s main advantage and why it’s economic is that it’s quick. It can do assigned tasks to whatever degree tech like writing software, graphics, video, etc. allow it to do so. AI can work to instructions, often pretty vague ones in some cases.
What’s happening with the doom and gloom at the moment is that a limited understanding of tech is now doing astrological predictions at best. Some are carefully not predicting what this tech will actually do, but the guesses are obvious.
To put it another way – The AI pundits are working with a piece of string that doesn’t exist yet.
Consider:
- AI sources information from a range of sources using searches. Everyone knows how efficient searches are these days. These searches are the epitome of laziness, often worse than in the 90s.
- They’re searching for things often written to commercial SEO and SEM specifications. A benchmark level of 5% of all data entry is in some way incorrect. The quality of information is often appalling and outdated. This is a reliable base for what, exactly?
- AI is totally dependent on other technologies to work at all. That’s hardly an advantage. It’s a weakness. You can also use those same technologies to deconstruct anything it does and see how AI did things. You could out-AI the AI simply by adding variants to the originals.
- The chatbot interface is a very useful feature, perhaps more useful than anything else at the moment. Quick responses are time-savers.
- AI works according to assigned tasks based on instructions from the interface. It doesn’t initiate or invent tasks. Any AI tasked by an idiot or more likely management team of idiots is likely to generate some dumb stuff.
Now, the more exotic business-level issues:
How do you fit AI into physical business systems and workflows? Why? To achieve what? What are the values? These are anything but rhetorical questions. You could spend millions on something you may or may not need.
Will it work with blockchain? It probably will have to do that. This is one of AI’s strengths, but try applying that to NYSE or other high-volume transactions.
How secure is AI? Can it be hacked and exploited? Probably. Are we hearing a word about that? Not really.
Who are the drivers and decision-makers in a business? People. Not AIs. There’s a reason for that.
Can AI manage “dashboard” businesses? Yes. Can it do sales? Only if very well scripted, and that scripting process is likely to be clumsy and highly competitive. The mix of clumsiness and competitive issues shouldn’t be too much of a mystery.
Anything called “analysis”, be it markets, performance, or whatever, requires a very strong, often specialized knowledge base. Can AI do that? Are you sure? If you’re not sure, that could get messy.
Read the Euronews link and check out the jobs affected by the expected impacts of AI. Now consider the practical realities of those jobs and what could go wrong, because it definitely will.
Next, ask your AI to write a report on something. Read the report. Would you put your name on it? Does it deliver the critical information it’s supposed to deliver? Is the report actually useful, or a shopping list?
Arts and media – How “creative” is your AI? Does it introduce scarring prose or devastating imagery? How competitive is it?
Remember – All of this will be obsolete next year at the latest. There will be an AI bubble in investment. There will be an AI bubble in consumer usage. You want to throw billions at that? Maybe not.
At the moment, AI is fetal. When it can act independently, it will be real intelligence. Until then, it’s scripted and subject to the limitations of its environment. Unless used intelligently, AI can’t be truly “intelligent”.
Hope is based on forward thinking. Something good will happen. Can AI deliver on that?
____________________________________________________________
Disclaimer
The opinions expressed in this Op-Ed are those of the author. They do not purport to reflect the opinions or views of the Digital Journal or its members.
