AI's deflationary effect, the question that could save your job, NVIDIA blows everyone away, AI comes for nursing, ChatGPT 5 coming soon, and more...

Thought of the week: the deflationary effect of AI

People often seem surprised that translation services still provide a substantial part of our company revenue. For many, translation was a tricky problem that technology has largely solved, so it should be generally available and almost free. They almost seem shocked when I explain that we have an in-house team of writers and expert linguists whose job it is to write, edit and translate important texts, such as flagship UN publications on climate change. To help them understand why humans are still better than machines at content generation that matters, I use the analogy of fast food vs fine dining. Like fast food, machine translation is good for quickly satisfying the hunger for localised and fairly ephemeral content, but if you want your audience to enjoy, remember and truly enjoy what they consume, then you still need human experts in the loop.

But while most clients understand the inherent trade-offs between cheap and fast vs slow and expensive, one of AI’s most insidious effects is the way it distorts the holy trinity of project delivery: cost, quality and time. AI creates the perception that you can actually have it pretty good, almost instantly and for peanuts. Ultimately AI, and automation in general, exerts a deflationary (or downward) price pressure that is only going in one direction. From translation to tutoring, customers will expect things to be fast, dirt cheap and of acceptable quality.

However, while prices will fall for average service delivery, any company that goes above and beyond average will thrive, so be human and outcompete mediocrity by over-delivering.

Are you 'super' or 'superfluous'? Answering this question will determine whether you keep your job

I remember getting canned from my management consultancy job almost 16 years ago, and though it felt like a punch in the stomach at the time, I still remember the sage words of a senior partner who trotted out the cliché that it would be the “the best thing that ever happened to me”. As much as I loved working with clients such as BP and the Foreign Office, I was not so super at upselling colleagues on to UK civil service projects at a rate of £1500/day, so I became superfluous.

I was reminded of this when I read an article in the Financial Times about what roles AI will make superfluous, but its central thesis about adding value to your company/customer extends beyond the 'AI will steal your job' narrative.

If you run a company or manage people and are assessing staffing levels, the key question will increasingly be: 'Is this person super or superfluous?' Super staff are those that make your heart skip a beat if they knock at your door (or, as is increasingly the case, email you) to tell you that they are leaving. They are the ones that have particular talents, like binding teams together and taking the initiative with clients, or, when managing projects, just being a positive soul that makes coming into the office worth the trek. These are the people that can’t (or shouldn’t) be automated away.

Superfluous staff are the quiet quitters who grumble about a company on social media and who are ultimately in the wrong job. This may sound overly harsh and, let's be honest, many people don’t have the luxury of choosing their employer. But for these people, being found to be superfluous can be a good thing. If staff feel superfluous, it's an opportunity for them to take ownership of their careers, retrain and do something they love rather than quiet quitting.

Why Jensen Huang (the head of the $2 trillion chip company NVIDIA) is the new Taylor Swift

The key AI event last week was NVIDIA’s annual GTC conference. NVIDIA makes the graphics cards (GPUs) that power the majority of AI applications, including ChatGPT. Hype aside, NVIDIA is the most important company in AI, and its $2 trillion valuation means that it's extending beyond just providing the engine that drives the AI car to also providing the platform for a future world where AI enhances every aspect of human life and industry. At the Taylor Swift concert-like conference, Jensen Huang showed off everything from ChatGPT for drug discovery to Project Gr00t, NVIDIA's new platform for developing humanoid robotics. Watch this video to see where physical AI is headed in the near future.

AI is coming for nursing!

NVIDIA also announced that it's partnering with Hippocratic, a US company that shamelessly brags about how it can undercut real human nurses with its cheap AI agents that offer medical advice to patients over video calls in real time.

According to the marketing guff, Hippocratic claims that its AI nurses outperform human nurses in bedside manner and education, and narrowly miss on satisfaction, in line with a survey.

This is the kind of stuff where I think the ‘tech bros’ don’t really understand the critical and physical work nurses do (and how poorly paid they are for the privilege). In a world where we are living longer and health systems are straining to provide even a basic level of care, AI can play a critical role in letting us do more with the same resources rather than less.

Why the developing world loves AI

A recent survey of workers around the world (sadly missing out Africa :-( ) found that those in Asia have a significantly more positive outlook on the impact of GenAI tools on workplace productivity compared to their Western counterparts.

Well, of course they do! By using ChatGPT on a mobile device, a worker in Sudan now has access to exactly the same AI ‘intelligence’ as someone who sits in Stockholm. A Swedish office worker clearly has way more to lose from AI upending the cosy status quo than an African, so it's no surprise that developed country angst contrasts with developing country optimism. AI really does, at least for now, level the content-generation and low-to-medium level analysis playing field.

But this levelling will only be short-lived as developed economies bring back a raft of services that have been outsourced to armies of low-cost overseas workers. From customer service to data entry, these lower-skilled administrative services can now be automated for peanuts, which will remove a key revenue stream for many of these countries.

Aside from a small income stream, these low-skilled jobs add little value to developing countries, so the focus should be on upskilling.

The new version of ChatGPT is coming out this year and it’s going to be awesome

I’m a great fan of Sam Altman’s ability to constantly surprise and over-deliver, so I encourage you to watch this video. In it, he talks candidly about Sora and the drama surrounding his ousting and reinstatement at OpenAI. Moreover, he addresses what everyone wants to know: When is ChatGPT 5 going to come out?

The answer is by the end of the year, that it will be “an amazing model” and that, before this, various other features will be launched over the next few months. He’s also very bullish on when he thinks artificial general intelligence will be achieved.

Other interesting AI news this week

• Is AI becoming the new 12th man in football? See this article on how Liverpool is using AI to skew the odds.

• The national AI race is heating up, with India, Denmark and Saudi Arabia all announcing plans to spend huge sums on sovereign AI systems. No word as yet from any African countries.

Tools that we most certainly are NOT playing with this week

Talk To Your Ex allows users to import their exes' texts to keep chatting and dating, even after being dumped.

BypassGPT generates AI content that is undetectable by AI detection.

BadGPT is an unhinged and uncensored AI that will answer any question.

That's all for this week. Subscribe for the latest innovations and developments with AI.

So you don't miss a thing, follow our Instagram and X pages for more creative content and insights into our work and what we do.