The AI of the future. Is it a future that we want?

26 6 minutes read

The AI of the future. Is it a future that we want?

Listen to this article

It seems inevitable that most of us will have our workplaces impacted by AI. Whether or not this impact will be overall positive, or negative is to be seen. I read with curiosity that at a time when Dubai based logistics giant DP World is looking to roll out AI and Automation across Australia’s ports, that the Maritime Union of Australia (MUA) is demanding to have their members hours reduced to 28, but maintain the same pay for working a full week as a result. How reasonable is this for the ports that employ of MUA workers to pay “full whack” for effectively a half working week? I would be interested to hear what the upside is for Australian ports, or those consumers who pay the pull-through costs of goods that transit Australian ports? It would be doubly curious to hear from the MUA what they believe the impact on overall productivity is expected to be should this 28-hour week be accepted.

Reference BBC article, 08 July: Australia dock workers call for 28-hour week in AI talks

Whilst I am not usually one to side with unions; the MUA statement that “(AI Technology) … should be used to improve workers’ lives, not destroy them” is one that we should all reflect upon. The flip side of this statement is to consider recorded productivity levels of those MUA members working in Australian ports, and their overall value to this industry. For a union that describes itself as “militant”, it would be interesting to hear the MUA’s views on why Australian ports lag behind most international benchmarks for productivity; and why the MUA is often accused of “restrictive workplace arrangements, protracted industrial disputes, and slower ship turnaround times”. Such views seem reflected in the Australian Government Productivity Commission reporting; for example:

Reference: Inquiry Report – Australia’s Maritime Logistics System | Productivity Commission

Those of us whom lived in Australia during the 1980’s would probably remember the MUA driven 1998 waterfront dispute that occurred during the Howard federal government, and the damage that caused to businesses and the economy. Reference: The 1998 Waterfront Dispute | Archives

Given the “productivity challenges” the Australian ports face from MUA, and the accusation of apparent costs of “paying off” the MUA to avoid industrial action (Reference: Trade union royal commission: MUA given large payments to prevent industrial action, inquiry told – ABC News, then it is little surprise that DP World would seek to streamline it’s port logistics operations, and that it would try to do so without in-depth consultations with the MUA. We wait and see whether the MUA looks to embrace the inevitable future, or seeks to fight is as it did in 1988. Perhaps one option DP World could consider in rolling out productivity and in streamlining it’s port operations, is to pay MUA workers to not turn up to work at all? Would “zero hours” and an entirely automated process result in greater overall productivity than 28 hours from MUA workers? That would be an interesting question. I would expect that most consumers and users of the Australian waterfronts would lean towards seeing MUA workers gone and replaced by AI as being more of a good thing than a bad one. As with so many things in life, people should be careful what it is that they wish for.

The vast majority of us working outside of Academia in technical Engineering execution aren’t union aligned, nor do/does our various professional institutions play much of an active role in our workplace terms and conditions. Many of us who work as consultants or as “Limited company contractors” are pretty much on our own when it comes to negotiating working contracts and working conditions. Our only leverage is to effectively “take what is offered”, or leave and seek work elsewhere. Any engineers who consider themselves to be indispensable might want to consider how much of a hole is left behind when one takes their hand out of a bucket filled with water?? No doubt some do work better and more effectively/efficiently than others, and believe themselves to have unique skillsets. Sadly, I am not sure that is all that important to those employing professional engineers, or seeking to replace those professional engineers already employed with alternatives. I would also challenge that the vast majority of us Professional Engineers have skillsets that others could do if they really had to. Perhaps not as well, but those tasks we are set would eventually get done by someone else. Given many engineering design houses are already employing “value execution centres” in regions with significantly lower living costs to Western Europe, then this transition is well underway (and has been so for some time). I would expect that for the most, AI will initially be a (perceived as) lower-cost alternative to those “value execution centres” already in operation. The other side to this coin is whether currently free and low-cost AI services will remain so in the future. Construction and running costs of AI datacentres isn’t small, and those same centres come with a heavy environmental impact (Reference: Energy demand from AI – Energy and AI – Analysis – IEA).

What happens when the owners/operators of these datacentres start pushing those costs on to the consumers of AI? Once society and (should) engineering design becomes reliant upon AI, it isn’t unforeseeable that AI companies will monetise usage of this service and thus start escalating usage costs and fees. That future has already been told by outgoing Apple CEO, Tim Cook in “hinting” at upcharges to future versions of Apple’s voice assistant, Siri AI. Reference: https://www.cnet.com/tech/services-and-software/apples-ceo-hints-that-heavy-use-of-siri-ai-could-cost-you/ ).

Whether AI ends up as actually a lower cost, and this lower cost is actually better value is to be seen. In engineering, the devil is often in the detail. Most wouldn’t notice 100 things designed right; but one error has the potential to be catastrophic. What happens when AI makes a wrong decision, or goes rouge as it has recently when/where is hacked into client systems at technology firm Anthropic (Reference: Anthropic’s Claude AI escapes tests to hack three organisations), and only a few day’s earlier at Open AI (OpenAI says its rogue AI tried to hack other companies ? Surely, we can’t expect AI to work in our best interests, and be “ethical” when so much of the information AI may draw “human ethical behaviour” from doesn’t show humans act in this way. If we let AI into our workplaces and workspaces then how far in does it get? I guess the question as highlighted by firms Anthropic and Open AI is “how far CAN it get”? If we assume that we can build ever more complex sandboxes to contain AI, then is it not worth consider that AI may find ever more complex ways to defeat that containment?

Once engineering entities, and society in general is reliant on AI, then what stops AI providers increasing the costs of accessing this same AI to a point companies reduce their human work force ever more to cover that cost. With more AI reliance, and less human interaction in Engineering design, then who takes ownership of this design and how (and who) is it checked by? The other aspect of this push into AI is how we stop malicious acts given we seemingly can’t stop rouge acts by AI (apparently) under supervision? As engineering design and execution becomes ever more reliant upon AI, then it stands to reason that the more recent graduates will also become more reliant upon AI, especially for the “donkey work” most of us were assigned when we first entered industry. It could even be that there will be no work for them in engineering design and execution period! Here, the MUA statement is worth repeating in that “(AI Technology) … should be used to improve workers’ lives, not destroy them”. But it already is!

How do recent Engineering graduates learn when in Industry and develop their skillset? From the knowledge and experience of senior engineers (whom might also be phased out by AI), or from AI generated design and execution? If AI ends up being the case for design and execution, then what function would or could recent Engineering graduates serve that AI can’t?