AI. Brilliant but flawed - just like humans

I was recently reminded of the Lassonde Curve which illustrates the value shift during the lifecycle of a mineral discovery. 

Source: Small Cap Investor: Understanding the mining life cycle

Coincidentally, whilst working on a consulting project in biotech currently, I was made aware of a similar curve in biotech which describes the phenomenon of escaping the valley of death.  We see this in tech more generally. The overt similarities interested me. 

As we often do these days, instead of Googling, I asked claude.ai: Do we see the equivalent of the Lassonde curve in biotech?  Mostly I just wanted to find a source for the biotech curve. 

Claude’s answer surprised me on two counts. 

Firstly, claude.ai was unaware of the Lassonde Curve.  Realy? A simple Google search will reveal the curve above, along with plenty of references.  At least claude.ai was not wrong, just useless. 

I subsequently found the valley of death in the biotech sector also via a simple Google search.

Source: Pharmexec: Escaping the valley of death

The resemblance to the Lassonde curve is obvious.  That it is so should be unsurprising, but it was new to me in that moment. But what came next surprised me even more. 

Claude went on unprompted to tell me:

“However, if you’re asking about learning curves or scale economies in biotech, there are definitely some interesting patterns:
Cost-per-discovery curves: As biotech platforms mature, the cost per successful drug candidate or diagnostic often decreases due to:
• Accumulated data and insights
• Refined screening processes
• Better target identification
• Platform efficiencies
”

— claude.ai (19 Sep 2025)

I knew that this is wrong. I use a drug discovery curve in a session I teach on strategic innovation in the MBA which demonstrates the exact opposite is true. This reality is true across multiple sectors: our research efficiency is declining. This is counter to the assumption many people hold. The explanation has to do with ‘combinatorial innovation’ but that’s a whole other blog.

Declining Research Efficiency.
Source: Scannell, J.W., Blanckley, A., Boldon, H., and Warrington, B. “Diagnosing the Decline in Pharmaceutical R&D Efficiency,” Nature Reviews Drug Discovery, 2012, 191-192.

So, then I challenged claude.ai: Your comment on capital efficiency curves is at odds with the empirical data ... do you want to review your comments and reassess. 

If you have any experience with these AI tools, you know what happened next.  Claude suddenly found it agreed with me: viz.

“The empirical data actually shows that capital efficiency in biotech tends to worsen over time, not improve. This is the opposite of what I suggested … The biotech industry seems to exhibit something closer to an inverse learning curve - where accumulated knowledge reveals how much more difficult the remaining challenges are, rather than making them easier to solve.
This is fundamentally different from manufacturing industries where experience and scale typically drive unit costs down.
”

— claude.ai (19 Sep 2025)

What is the lesson we should draw from this?  It is obvious.  For all the surprising skills of AI to deliver insights and write cogent narratives, often with more skill than many of us might muster, you simply cannot rely on it to give you advice in domains where you do not have domain expertise.  It will take you down a garden path and leave you stranded and embarrassed. 

This should not be news to anyone, but it seems some missed the memo:

I don’t particularly want to call out Deloitte on this - I am certain they are not orphans in this - but the AFR headline so clearly made the point. It is worth reading the article to appreciate the scale and scope of errors. And if you’ve spent any time using AI you would think the risks were obvious.

In the MBA strategy & consulting unit I teach at UWA, where students undertake a consulting project for ‘clients,’ I tell them to use AI - they would anyway, but let’s make it overt. But I also require them to develop a policy position on the use of AI.   Some of the teams did a really good job on the policy, and if they follow it, they won’t find themselves embarrassed.

In the meantime, remember:

AI is brilliant, but like many brilliant people, highly flawed. 

The Board/Executive strategy workshop - a practise note

Henry Mintzberg wrote 25 years ago: we need better practice, not neater theories.  Amen to that.  Much of the writing on the ‘formal strategy process’ ignore the practical reality that strategies have a life cycle.  Twenty years ago Gary Hamel observed strategies decay for four reasons: replication; supplantation; exhaustion; evisceration (but that’s another blog).  The point is, strategies have a life cycle, and the process has to reflect that context. 

This blog highlights some elements of a recent Board/Executive strategy workshop, with about 20 people in the room (it is a large board).  It followed a deep, fundamental, strategic review and workshop I led about 12 months prior in a workshop that involved the board, the CEO and one other executive. 

Thus, the purpose of this recent workshop was not a re-examination of the strategy: the strategic direction was clear and agreed.  A few things had changed in the external environment since the prior workshop.  Whilst this had caused the sequencing and prioritisation of some of the strategic initiatives to be adjusted, it wasn’t a case of a wholesale review.  With a nod to the former UK Prime Minister Harold Macmillan: what was the greatest strategic challenge of the CEO – “events dear boy, events”.

The purpose of this workshop was a chance for the board to hear from the executive on the implementation program, check alignment, and provide feedback to the executive. 

My bias in any strategy process is toward conversations.  Great strategies are rarely discovered primarily through analytics (thank you Jeanne Liedtka), but through conversation, action and reflection.  The quality of the conversations is a one indicator of the likely success of your strategy – not just in ‘design’ but in execution.   

Given this brief, what were some of the design features?

§  Early engagement with the executive to shape the content and messaging.  What does the board need to know[1]?  What do you want the conversation to be about?    When you are the executive immersed deeply in the content, it is sometimes hard to lift yourself out of the detail. 

§  Be clear about what you want from the Board?  Sometimes you just need them to be aware of strategic context.  But sometimes you need them to make a call, or at least give you some directional guidance. Are they happy with the pace and direction?  Design with intent. 

§  Bring the board into the conversation early.  We began with a simple question: how do you feel as we start our day?  This brought some real energy into the room.  And there was strong alignment from the outset – more than I anticipated. 

§  Drawing out the input from across the board.  With a large board, it is too easy for some to sit back.  You need engage individuals, directly invite them to contribute.  Sometimes you want to provoke.  My role isn’t to simply make sure everyone is being polite.  I want to know the issues are being surfaced, and challenging views are given voice. 

§  Synthesis.  Often as facilitator we add value by synthesising the conversations as they progress.   At this workshop I particularly wanted to board to ‘own’ the outcomes of the conversation.  And so at the end of the day I called on a few of the key influencers to offer their synthesis.  At times I would build on their comments to connect the different threads. 

In the wrap up, many of the board members commented on the strength of alignment.  Clarity of direction – alignment – is more than many organisations achieve.  As Richard Rumelt [Good Strategy, Bad Strategy] observed: 

“an organisation’s greatest challenge may not be external threats or opportunities, but instead the effects of entropy [a measure of the degree of disorder in a system] and inertia”

One of the more vocal and challenging board members commented: “I’m even more energised than I was this morning; it was a great result”.  That’s a win!

[1] If you’re not familiar with it, check out Minto’s pyramid.

Slowly, then quickly ...

In mid-2015 Woolworths was the poster child for a board and executive that forgot managing a company is about more than shareholder returns.  Today that moniker belongs to QANTAS.  To quote Joe Aston – a fierce critic of QANTAS and more specifically, Allan Joyce – and Ayesha de Kretser in the Financial Review (Saturday 9 Sept):

“New CEO Vanessa Hudson takes over an airline that makes more money than ever but is the most hated brand in the country.”

It is no small feat to go from one of Australia’s most iconic brands to most hated.  How did it get to this?  To borrow from Hemingway: slowly, and then quickly.  But it gave me a reason dust off and refresh a blog from the archives. 

What does the Woolworths experience teach us? 

Woolworths held their margins too high for too long, prioritising shareholders over customers.  And customers voted with their feet – and their wallet.  Woolworths share price dropped 30% over 12 months [40% over 24 months] from its mid-2015 peak.  Commentators expected it would take at least 2 years to recover a reasonable sales growth trajectory. 

In February 2016 Woolworths’ new CEO announced his first priority was ‘to focus our team on customers’.  Woolworths reduced prices 3.8% in Q4 2016 [1].  This had immediate effect, with their customer satisfaction jumping from sub 70% to 75% in the final quarter of FY2016.  Woolworths share price recovered by October 2019.  This was also the time of the Masters debacle (see here for that story) so there were confounding factors.   

The primacy of shareholder value is much loved by many, and the implications of not delivering on shareholder value are ever present. The market can be pretty harsh when the returns are not ‘satisfactory’.

But near-term shareholder returns are no proxy for organisational health and long-term value.  Many highly respected business leaders argue against the cult of shareholder value. Counter-intuitively, Rosabeth Moss Kanter has argued:

“The companies that are most successful at maximising shareholder value over time are those that aim toward goals other than maximising shareholder value”

In 2010 Roger Martin, the Dean of the Rotman School of Management argued that maximising shareholder value “is a tragically flawed premise, and it’s time we abandoned it … and made the shift to customer capitalism”.

Peter Drucker argued that “the sole purpose of business is to attract and retain customers”.  While most executives believe the ‘accepted wisdom’ in the US is that businesses are obliged to maximise shareholder value, when asked their own views the highest rated corporate purpose was ‘to serve customers interests’.[2]  It’s not that Drucker et al. imagined companies did not need to make money. Failure to make an economic return is ultimately fatal.  But your economic model will fail without customers valuing what you offer – and feeling valued.   

What about the talent equation?  Investments include both financial and ‘human capital’ (our talent). Financial capital without talent is just money.  Just as the financial investors expect a reasonable return, so too, our human capital should expect a reasonable return on their investment.

If you see people as simply an instrumental input to ‘production’ then you limit what’s possible. This is the fundamental difference between transactional leadership – which produces what you contract for – and transformational leadership – which produces more than you expected.

Some very successful CEO’s actually live by this principle:

§  Business has to give people enriching, rewarding lives, or it’s simply not worth doing … Richard Branson

§  You have to treat your employees like your customers … Herb Kelleher (CEO Southwest Airlines)

Thus, businesses compete in three markets[3]: capital; customers; and talent. Arguing which of these three stakeholders has ‘primacy’ misses the point: to succeed, a business must succeed in all three markets.

View fullsize

In an ideal world, these stakeholders sit in a state of constructive dissatisfaction.  Shareholders seeking more profit; customers more value; employees a better deal.  It is constructive dissatisfaction because it is this tension between ‘what is’ and ‘what could be’ that is the animating force of change. 

Is it too glib to suggest that a business can manage this system so that these forces are not in a state of tension?  Wouldn’t that be the equivalent of having zero debt on the balance sheet:  a lazy balance sheet?  Nothing more being demanded by anyone?  This is arguably a recipe for failure as much as getting the balance wrong.    

For too long now QANTAS has given primacy to shareholders over customers and employees.  Fill in your own QANTAS horror story.  Accountability for this sits as much with the board as it does with the executive.  Watch this space.    

What are the lessons for your business?

Create space for an open conversation in the board room and within the executive team.  Is there a felt level of ‘constructive dissatisfaction’?  What measures might suggest the tension system is out of balance?  Do you allow – invite in – conversations that challenge the current balance?  Do you listen to outside voices?

… the unfashionable strategist

PS: check out The Fin podcast: ‘The real story behind Qantas’ rapid descent – and what’s next’.

[1] Excluding tobacco

[2] Unpacking Corporate Purpose.  A report on the beliefs of executives, investors and scholars (2014). www.aspeninstitute.org/UnpackCorporatePurpose

[3] You could make an argument that the model should include suppliers, and potentially broader stakeholders, but that’s for another blog. 

Understanding VUCA: a strategist's guide

In the world of strategy these days people often toss around the acronym #VUCA: volatility; uncertainty; complexity; ambiguity. But can you tell the difference between each? And what does it mean for your choice of strategy tools. I put together a one page guide (click here) which describes each; provides a practical example from the world of business; and suggests a few responses to each. 

Good luck. Feel free to share …

DDB … a strategist’s view

Smartest guys in the room … what could go wrong?

What could go wrong?  Plenty it seems.

Commonwealth Bank (‘Which bank?’) is just the most recent example of overconfidence in the ‘intellectual capacity’ of a leadership team to steer the organisation through the ‘white water’ of modern corporate life.  In its recent report on failures at CBA APRA reportedly found:

• The board relied too heavily on senior management’s ‘high IQ’ and their ability to take care of all things
• That their risk management processes relied too heavily on ‘raw intellect over comprehensive analysis’

This is a pattern that we see repeated too often and yet we still fall for it.  We’re a bit like Charlie Brown and Lucy (in the old Peanuts cartoon).  Despite Lucy repeatedly pulling the ball as Charlie goes to kick it, each time Charlie imagines 'this time it is different'. 

View fullsize

There’s been a host of studies on team performance and team dynamics over the years that show that teams full of ‘really smart people’ don’t perform as well as teams which might have 1-2 really smart’ players (and not necessarily the CEO).  But somehow, we too often assuming that the answer to complex problems is ‘smart people’.    

This has proven to be a very expensive assumption. 

Perhaps the doozy of them all was Long Term Capital Management.  LTCM was a hedge fund that boasted two Nobel laureates (Merton & Scholes, famous for options valuation theory), multiple PhD’s, and fund management veterans.  LTCM grew to over US100B in funds within 4 years, showing 40% annual returns.  But for all that, the Asian financial crisis triggered a collapse that saw LTCM facing US1Trillion in default risks.  Ultimately the US Treasury had to step in to stablise the global financial system.   So much for genius. 

A close second: Enron.  From a pipeline company in the 1980’s, Enron grew into the world’s largest energy trader.  The collapse of Enron is told in a documentary ‘The Smartest Guys in the Room’.  It traces the rise and fall of Enron, which grew to be the darling of Wall St, until it collapsed.  The Chairman and Chief Executive of Enron at the time were Kenneth Lay and Jeff Skilling.  They have been described[1] as ‘two supremely arrogant and belligerent men who believed they were the smartest guys in the room: that through sheer cleverness and creativity they had brought into being the most innovative corporation in the US’.  Indeed, Enron was named ‘America’s most innovative company’ by Fortune for six consecutive years.   The collapse was triggered, at least in part, by investigative reporter Bethany McLean who was undistracted by the glossy brochures and glitzy premises.  It seems she didn't drink the Kool Aid.  

As the dominoes began to fall a pattern of appalling behaviour emerged.  For example, when the power markets were deregulated in California, Enron shut down energy generation to create power shortages and drive up energy prices.  According to The New York Times, the top brass at Enron realised what was happening ‘but like a mad and dysfunctional cult, everyone carried on’.  This description has parallels to the revelations of the current banking royal commission.

My final example is closer to home.  The CEO of one of Australia’s leading companies had a reputation for his prodigious intellect.  The company prided itself on capturing the best talent available in the sector.  Despite this, with his powerful intellect, he developed a reputation for intellectual bullying.  But in the years prior to his retirement he was persuaded to introduce a values-based leadership development program.  In a post-retirement interview he observed, with a sense of profound insight, that he had recently come to the realisation that ‘it was all about the people’.  One got the impression the interviewer was nodding approvingly at such wisdom. 

I was bemused.   How is it possible that the CEO of a major company, with a reputation for a prodigious intellect, took until his late 50’s to realise ‘people matter’?  This is Management 101 on any MBA.  Tom Peters, management guru, author and former McKinsey partner, has been telling us this for as long as I can remember. 

The simple reality is that intellect can only get you so far.  Perhaps intellect is a bit like money: more is better up until you reach a point at which the upside plateaus.  At this point, other factors – wisdom, emotional intelligence, and the collective intelligence – begin to play a much bigger role.

[1] Oppel, R. A. and A. R. Sorkin (2001). Enron's collapse: The overview.  Enron collapses as suitor cancels plan for merger. The New York Times.