A marketers explanation of how ChatGPT works.

A marketers explanation of how ChatGPT works.

ChatGPT has blasted into our consciousness over the last 2 months. It has created an equal measure of excitement as people see the opportunities for leveraging their capabilities, and dismay at the problems they see being created.

Both are right, but if we are to make judgements about which side of that fence we choose to sit, it makes sense to understand a little bit about how it works.

These AI tools work on letters, and groups of letters, which then make up words, and the probability of one letter following another, and then another, and then one word following another, and another.

There are about 40,000 commonly used words in English, and billions of words published. From this database computation can give you the probability of a letter following another, eg. The probability of a U following a Q is very high, the probability of a V Following an L is low. This probability logic is extended to groups of 3, 4, 5 letters, one calculation of probability at a time. The outcomes of those cascading probability calculations transforms letters into groups that make up words based on the text used to ‘train’ the software.

Many words have multiple meanings, depending on the context in which it is used, homonyms. Sometimes the spelling is different, but they sound exactly the same. We understand what is meant by the context in which the word appears. For example: if I said, ‘I am on leave’ everyone knows I am on holiday. By contrast if I said, ‘I am going to leave’, it means I am about to depart whatever event we were at. I might also leave something for you at the door.

The word ‘leave’ is spelt and pronounced exactly the same way every time, it is the context in which it is used that makes the difference.

The juxtaposition of words also makes a difference to our understanding. If you remember your primary school grammar, it is all about the position of the subject and the verb.

If I was to say: ‘I am going to leave the party‘ the subject, object, and the verb are in the correct position in English for easy understanding. If I was to say ‘the party I am going to leave‘, most would understand, but would be expecting me to say more, despite the words being identical, it is just the position that changed.

Linguists have studied these relationships for years. Their mantra is: You will understand a word by the company it keeps.

If you take this to its logical extreme, the position of every word in a body of text has an impact on the understanding of every other word, and group of words in the same body.

If the surrounding text to my sentence is about going to a friend’s place for a drink, that will lead to a probability that the ‘party’ has to do with a social event. On the other hand, if the surrounding words were about politics, the phrase ‘I am leaving the party’ takes on a completely different meaning. All these considerations are taken into account by the magic of the probability of me leaving the party when the words friends and drinks are in the surrounding copy. Should those surrounding words be government, and policy, it is more likely the party I am leaving would be a political one.

The operating system of Open AI, and others, have scraped the web for all text published, and stuck it into what amounts to a huge multidimensional spreadsheet. The machine calculates the probability of any one letter appearing after another, then any word appearing next to another based on the occurrences of those letters and words and groups of letters and words in the scraped text. It does this over and over again, spreading the web of probabilities of words and groups of words appearing together, in a particular order, wider and wider, one word at a time, across the body of copy.

This process is extraordinarily computationally intensive. It is hugely expensive to build and program machines that can do these enormous sets of calculations on this amount of text.

If you give such programs a general brief, the best it can do is return a general response. The more detailed you can make the brief, the more explicit the context, the better the machine will be able to use probability to find that combination of words that best matches your requirements, then spit out a response to you.

As a marketer, you understand that when giving a creative brief to an ad agency, the more detail you can give the creatives, the more relevant will be the creative responses. A general brief will give you lots of ordinary creative responses. By contrast, a detailed brief that clearly articulates the target market, product benefits, and the value to be derived from the products use, will generate better creative responses.

ChatGPT is no different, so for good results, give it a good brief.

What makes this so powerful for those who are expert in their domains, is that they will be able to give better briefs, and so have returned better results, which will then be the basis of their creative thinking. This offers the opportunity to improve on the best that has been done to date. For those who are not as expert, their briefs will not be as good, the context in which the machine defines probabilities will be wider, so the output more general, generic, average, and average these days increasingly simply does not cut it.

I hope that helps.

For a more detailed and technical explanation of how ChatGPT works written by an expert, go to the fifth PS at the end of this blog post published when I first stumbled across ChatGPT in December last year.

Header Credit: Dall-E. The brief was ‘ChatGPT algorithms working hard to compute copy in a surreal setting’

The uncertain future of work and jobs.

The uncertain future of work and jobs.

 

 

Hemmingway observed in ‘The sun also rises’ that ‘the future comes slowly, then all at once’.

He has been proven right many times.

Since the release in November last year, ChatGPT has proven the future of AI is here, all at once.

That reality leads to the key question: so, what now?

We often look back on the spread of electrification as a template for thinking about the digitisation of our economies. It is a fair representation except for one small detail, which makes all the difference.

Electrification was a process that proceeded sequentially, piece by piece added as efficiency improved. From the beginning of the digital age, and the recognition of the reality of Moore’s Law, this has changed.

The driver of change has been compounding, each stage building on the previous, with increasing speed. While this has been seen by most as just normal improvement, the cumulative impact has been far greater.

Einstein noted that the most powerful force in the universe is compounding. Imagining the impact of compounding is really hard, makes my head hurt. To imagine it, there is still no better metaphor than the old rice on the chessboard fable.

The emperor promised someone (probably an ancient consultant) a payment in rice on a progressive scale, calculated as doubling for each of the 64 squares on a chessboard. 1, 2,4,8,16,32, and so on. It seemed like a good deal to the emperor who was clearly not mathematically minded.

By the 31st square, payment topped a billion grains of rice, enough to cover your average ancient town square. That is where the problems started as payment kept on doubling, quickly outstripping the total world production of rice.

The tipping point is somewhere around square 25, where the rice was a couple of wheelbarrows full, then seemingly suddenly, it became a vast amount.

Such has been the case with digitisation.

We have been watching its progression since Gordon Moore wrote his 1965 article predicting a doubling of the number of circuits on a single chip every 18 months. A bit like the emperor, we have watched and suddenly it seems we have reached a tipping point led by ChatGPT and its sibling DALL-E. Hot on Chats heels came ‘Bard’ from Google, although stumbling at the launch last week, and no doubt Amazon and Apple are close behind.

The difference we face to that faced by the emperor, is that had he used his abacus, he could have predicted the outcome of his agreement, as it is calculable, to a point. What happens now with the compounding of AI is not so predictable. What we do know is that it will be a disruptive force coming at us with compounding speed and power.

This power to increase the speed, accuracy, and therefore efficiency of the processes we digitise will extract a range of very high tolls. These will be the increased risk of personal data being available and almost inevitably used against us, amplification of bias, ever increasing complexity of the systems we will come to absolutely rely on but not understand how they do what they do, and a complete ‘rework’ of work. This revision of work will make the changes from the cottage industries pre industrial revolution look like minor adjustments by comparison, and will happen at lightning speed.

Of concern to me is that only a few have the scale necessary to ‘train’ these systems. Microsoft, Amazon, Google, and Apple have that scale, which will serve to entrench their dominance in the space. Theoretically governments also have the scale, but will be hobbled by concerns unshared by commercial players.

Within a decade, every current job, those that remain, will be almost unrecognisable, and there will be new jobs we cannot yet predict taking their place. What will remain is the human element of creativity, that capability that distinguishes human beings from all other species, the ability to do something completely new.

The good news is that we will still need engineers, architects, doctors, plumbers, and bricklayers, but the shape of their day will be nothing like it is today.

When digital photography took off, putting a quality camera in every pocket, most thought it was the end of photography as a profession. Not so. What became quickly obvious was that there was a clear distinction between the real, creative skills of the elite photographers, and those of the ordinary. The pareto distribution of photographic skill applied, and those that survived as professionals had more time and better tools with which to capture and express their images. This will be repeated in every job across the economy.

Unanswered is the question of how we educate our kids to thrive in a work environment we are unable to visualise.

Header credit: Dall-E. The instruction I gave Dall-E was ‘Surrealist impression of the change from cottage industry to knowledge work’ This was one of 12 generated in about 30 seconds. Look closely at the face.

 

The elusive formula for winning and managing government grants.

The elusive formula for winning and managing government grants.

 

There is considerable grant money being allocated to innovative solutions to technical and market challenges by all levels of government. Such a honey-pot attracts all sorts of characters with a whole range of motivations, along with the genuine applicants seeking help. In this environment, panels of disinterested departmental officials and sometimes so called ‘experts’ are called upon to make judgements. As has been demonstrated over the last few years, these judgements are not always followed closely when votes are in play.

Be prepared to acknowledge that there is a whole lot of ‘lottery’ involved. Judgements about your eligibility against a set of guidelines that can be ambiguous, convoluted, and occasionally contradictory, can be an enormously frustrating and time consuming exercise for applicants. In addition, despite what is said, innovation involves risk. No government wants risk, and bureaucrats are conditioned by their culture to be utterly risk averse. The most remote whiff of risk, an indication of potential failure which can be politically weaponised to end careers is abhorrent to project assessors, irrespective of the number of times the word ‘innovation’ appears in the literature and conversation.

Before you ever approach the process of committing the resources to apply for a grant, then managing it should you be successful, you need to understand 3 basic rules:

  1. Any grant funds will come into your P&L at the top line, so will add to profit assuming you make some, or reduce future tax losses. Most programs require cash co-investment, so make sure you discount the potential value of grant funds appropriately before you start.
  2. Notions of Commercial in Confidence, often a central driver of innovators is absolute poison to public authorities, whose whole mind-set is about levelling the playing field. Assertions of Commercial in Confidence, written or verbal are worthless, even when delivered in good faith, as the project proposal usually goes through multiple hands during assessment.
  3. To quote a senior bureaucrat during a conversation with me about the above two considerations: ‘when you get into bed with the government, who do you think is on top?” Recognise that grants come with strings, and managing pro-actively those strings, even when they seem somewhere between irrelevant and absurd, is essential to your ongoing sanity.

Assuming you have come to terms with these three factors and want to continue, following is a check list of what you simply must do, and not do.

Do’s

  • Ensure you have very clear objectives and project path before you set about filling in the forms. Adjusting your project plan, time frames, or objectives in order to meet program guidelines and make your application seem better, is a common and serious mistake. Ensure your project fits their guidelines perfectly, never adjust your project to fit. A bit of nipping and tucking may seem like it will enhance your chances, and it may, but most often it comes back to bite.
  • Clearly understand the objectives of the program. This sounds pretty obvious, and it is usually reasonably clear. However, there are always implicit objectives such as inclusion, equality, job generation, and most importantly re-election prospects that play an often unstated role.
  • Reflect back the words of the stated project objectives in your communications, and add in some that reflect positively on the implicit objectives.
  • Most programs work in rounds driven by dates. While this is often very inconvenient commercially, it better suits the bureaucracies. A project that is rejected in one round might be successful in another less populated by applicants, as the tendency is to break up the program funding into equal parts. So, persist. Ask for and take the advice on why your application failed this round, (‘the money ran out for this round’ will never be one of them, although it will often be the case) and work that advice into your application in the next round.
  • Be prepared to have some well academically qualified person without any relevant experience of your industry, and indeed life outside the bureaucratic bubble, believing they can and should give you strategic and operational advice. You will be well advised to politely acknowledge and follow this advice, at least superficially, if your application is to be favourably reviewed.
  • Always be prepared to report as per the schedules, preferably a day or two before the deadline. Be explicit in your application about the importance you place on these milestones and the attached KPI’s. These milestone reviews will always be a part of the grant contract, embrace them. Set about making auditing your project progress easy for the granting body.
  • When you are not successful with an application, try and find out why, so you can do better next time. This can be a hugely frustrating process, and rarely will you ever know for sure, as those trying to explain it will be paranoid about telling you anything that may be used against them. I once prepared a grant application for a regional manufacturing innovation program for a client, where the guidelines were an absolutely perfect fit. My client was located in a regional town, had two patents on parts of the process he proposed to use, so we appeared to ‘nail’ the innovation requirement, would have generated a number of jobs, and was value adding a waste agricultural product, but we missed out. I spent considerable time and energy trying to understand why, but failed. I ended up receiving a number of 4-page emails that were absolutely incomprehensible, and could not get through on the phone. The ‘official’ up to whom my questions and protestations had been pushed simply stonewalled me. Eventually, as I am sure was the desired departmental outcome, I and my client gave up to invest the time and energy in something useful.
  • Document everything, they will, and you might need to refer back at some point.
  • Ignore the preponderance of verbs and adjectives that will adorn the guidelines and accompanying material. They are simply a manifestation of the bureaucratic instinct to complicate everything, using 3 words when one would suffice.
  • Offer cream biscuits at the very least with the coffee in the unlikely event that they drag themselves out of the Canberra bubble and come to your offices. Lunch is better still, call it relationship building.

 

Don’ts

  • Do not get annoyed by constant insistence that you nominate the electorate and postcode where your project will take place.  Just give them something that serves as press release fodder, irrespective of how accurate it might be. Usually this will be your ‘head office’ even if there is absolutely no relevant activity beyond governance being conducted from that address.
  • Do not ever miss a deadline of any sort. When implementing a project, if it looks likely you might miss one, forewarn them, with the reasons, then, preferably, meet the deadline. The added effort to recover to the deadline will deliver brownie points. Any variation to the terms of a grant agreement are treated differently when they are a surprise, than when they are forewarned. This is really just common sense and courtesy, but I have seen tiny molehills blow up like Vesuvius in their absence. Such misses can motivate an audit. The right to audit will be written into the grant contract, but will probably never happen in the absence of some sort of catalyst that motivates action. When they do audit, they are usually ‘tick and flick’ exercises. However, noncompliance with the reporting schedule, or obvious inconsistencies that emerge from a cursory look can lead to deeper audits that are seeking to find the inevitable breaches of the guidelines and grant contract detail. Responding will be a time consuming, frustrating, and resource hungry exercise. You have things to do to move the project forward, and manage the rest of your business, while they have as an objective, finding out where you have cut a corner, adjusted priorities, or spent in a way that is even marginally inconsistent with the agreement.  Best to avoid that sort of scrutiny by overt compliance.
  • Don’t expect them to be as responsive as you expect. The sense of urgency you feel will have no effect on the pace of progress of your application. Don’t let it frustrate you, too much.
  • Do not counsel them on the challenges faced in filling in their demonic templated application forms. Somebody who may be commenting on your application designed it, thinks it is perfect, and might take such criticism personally. When they are difficult, as they normally are, ask for clarification, pointing out the deficiencies as inhibiting the quality of the information you are giving them, rather than pointing out their idiot template was generated by Satan.
  • Don’t become annoyed at the constant communication required by different people who ask the same questions as the previous incumbent. This is nothing compared to the changes in personnel that will occur during the project implementation. It will often feel like you were put on earth to train a seemingly endless stream of apprentices.
  • Never forget that most grant programs are competitive. Therefore, you are not only seeking to demonstrate to the assessors that your solution to challenges being addressed is worth supporting, but it is more worthwhile than any of the ‘competitive’ applications.
  • Don’t forget that those doing the assessing are just people, trying to do a job in a culture that will be entirely different to yours. Generally they do not set out to frustrate your ambitions, that is just an unintended consequence of the culture they must operate in, so do not overreact.

 

The benefits of grant funding.

  • Obviously, when appropriate, and well executed, the cash. Almost always this is the primary reason a grant is sought. However, it often becomes secondary to the following point.
  • Recognition, networks and the next grant. Governments live and die by the communication they generate, and networks they can leverage. Generally they are pretty good at it, having brought in communication professionals who do know their jobs. (I exclude advertising from this comment. Public servants generally know absolutely nothing about advertising effectiveness, but insist on their right as the client to dictate the ads, which is why there is so many wallpaper ads thrown at us) Once recognised as a compliant, PR friendly grant recipient, the networking opportunities are significant, and often prove to be the best outcome of a grant. Being a recipient, and having that good record of co-operation, gives you a head start the next time, as you are a known quantity, which reduces risk.

I hope that all helps, good luck, you might need it.

Header cartoon credit: Tom Gauld

 

 

 

Bing takes a sniff of (AI enhanced) Columbian marching powder.

Bing takes a sniff of (AI enhanced) Columbian marching powder.

 

Bing and its sibling ‘Edge’ have been coming third in a one-horse race for a long time now. Suddenly the emergence of the AI equivalent of a plutonium battery powered race whip in the form of ChatGPT has delivered a proper kick up the arse.

The world has changed, pivoted on a dime as they say, as a result.

No longer will Google search be the only game in town, and Chrome the default browser housed on 98% of devices. The new race has begun with a wider field, and no doubt some roughies hiding in the wings.

Microsoft announced 2 weeks ago that it has extended OpenAI’s models across their Azure services, widely used by developers, so who knows what might spring out of that.  Last week Microsoft confirmed ChatGPT is being incorporated into Bing and Edge.

Google have the most to lose here, so have scrambled to announce they intend to incorporate their version of OpenAI’s google-killer ‘Bard’ into search making it more ‘ChatGPT like’. It is just a pity the horse stumbled at the first hurdle by failing to answer a simple question, leading to a share price nose-dive into the turf.

This is a must win race for Google, as 80% of their revenue comes from advertising. With hindsight, they have bet the farm on the one horse, never a great strategy in a volatile environment.

It is going to be interesting!!

 

 

 

 

 

How will Google respond to the existential threat of AI powered search?

How will Google respond to the existential threat of AI powered search?

 

Never have I seen a more definitive example of Clayton Christianson’s ‘innovators Dilemma’ than what is being played out right now, in front of our eyes.

In summary, the dilemma is that dominating incumbent businesses are loathe to change the model that made them dominating incumbents. This results in them failing to innovate in ways that have potential to erode the cash flow from former successes.

Christianson had many examples in his book originally published in 1997, but none better than the existential crisis being faced by Google from ChatGPT, launched in November 2022.

Googles control of the search market is almost absolute, with a share of well over 90%. When you add in the rebranded search engines that simply use Google under another name, like Apples  Safari, and discount the mistakes that lead to Microsoft’s Bing being clicked, it is probably 97% or above.

Ask Google a question, and the first 5 or 6 responses are ads. They represent potential answers to your question, but just potential from the indexed websites. The revenue from those ads that also follow you around the web is 80% of Googles total revenue, most of the balance coming from ad revenue on YouTube. After scrolling through the ads, you will have to skim and review a number of possible sites that may deliver you the answer you are seeking.

Ask ChatGPT the same question, and you get back one answer. No ads, yet. You may have to become increasingly explicit in the question you ask, but the response time is close to real time, and you get the best answer available. It may not be the perfect answer, although we can expect it to improve, but it will save heaps of time.

Google claim to have a similar system sitting on the shelf. In addition, they made a $400 million investment in an AI start-up called Anthropic in late November, just after Chat was launched. I’m sure they have the capability to deliver an answer to Microsoft, as they have been playing with AI for a long time. Perhaps they did not launch because it is not yet perfect, what new product ever is, but more probably they delayed because it is a threat to the existing revenue of the business.

Since the early days, Google has sat on its mountain of cash and not innovated. They have fiddled at the edges, as shown by their site that keeps tabs on their hits and misses,  killedbygoogle.com but never confronted their cash cow, search, with any sort of  innovation that might eat their breakfast. This is in stark contrast to what Apple has been prepared to do, several times.

Whatever else happens, ChatGPT and its backer Microsoft have taken the initiative, and I suspect this will be the best $10 billion investment Microsoft has made in decades. Incorporating ChatGPT into Bing suddenly gives Bing a reason to exist and a competitive advantage to which many will be attracted.

I can only imagine there are late nights in Sundar Pichai’s  (Alphabet’s CEO) office currently as they try and figure out a way to combat this competitive threat while preserving their river of cash from advertising.

As I wrote this post, Google shares tanked and Microsoft announced a new generation of Bing running the next iteration of ChatGPT, customised for search.

Header: Google meets ChatGPT in the style of Monet in blogs used courtesy Dall-E, ChatGPT’s graphic AI stablemate.

Update No. 1. Feb 10, 3 hours after the original publication. probably the first of many.

I came across this Google post on their own site, via Visual Capitalist. If anything, it absolutely confirms the contention in the above post that Google have badly fumbled the ball. Timing is a much underrated quality in marketing. On several occasions, I have done the right thing at the wrong time, usually well before the market is ready, and failed as a result, only to see a competitor succeed at a later date.

Equity or loans: The entrepreneurs funding dilemma.

Equity or loans: The entrepreneurs funding dilemma.

Every start-up requires funding. A business plan, no matter how good, without cash for implementation will remain a dream.

Most start-ups get off the mark with investment of one sort or another by the ‘3 F’s’: Family, Friends and Fools, supplemented by savings from the aspiring entrepreneur.

At some point, after proof of concept when the aspiring entrepreneur needs more cash to fund the growth, or scaling of their now successful SME, they must seek alternative sources. The choices are simple: give equity in exchange for the funds, borrow them from some financial institution, or use a combination of both.

There are pros and cons to be considered for each path, the answer that best suits every instance will be different, and subject to all sorts of caveats and variables.

Pros for Loan funds.

      • The entrepreneur does not surrender any ownership and therefore control of the enterprise.
      • The entrepreneur gets to keep all the profits, assuming success. This assumes that the business is housed in a limited liability company, rather than on a personal basis, or even worse, a partnership.

Cons for Loan funds.

      • A start-up may (probably will) have trouble getting a loan at a commercially viable interest rate, as there is no or very little trading history. Cash is the ultimate commodity, and institutions only lend money when they are comfortable, they will get their money back, and/or the interest rate is such that the risk is deemed acceptable.
      • There is a debt that must be paid back with interest, irrespective of the success or failure of the enterprise. Interest however, is tax deductable, assuming there are profits that incur a tax liability.
      • A lender will often place restriction on how capital is to be used, require reporting, and demand privileges that do not add value to the enterprise.

Pros for Equity.

      • The entrepreneur does not have to pay back the capital, it is invested at the risk of the investor. The investor is absorbing the risk of failure into their equity calculations. If the business fails, they lose their money.
      • The absence of a schedule of repayments enables capital to be directed at the most productive uses
      • Those who provide equity are often in a position to offer more than just money. It is in their interests to leverage their networks and experience to benefit the enterprise in which they have invested. This advice can be of immense value, particularly to the young entrepreneur lacking the wisdom that comes with experience.

Cons for equity.

      • Equity is just another word for ownership. If you give some level of ownership to another party, that entitles them to a share of the profits as a function of their equity, when and if they emerge.
      • When the business is sold, acquired, or floated, the equity holders share in the proceeds, it is their opportunity to generate a return on their investment above the dividends received.
      • Equity also entitles a say in management, strategy, and management of the entity. This can be agreed, but there are regulations and accepted practice around the power of an equity holder, and in the case of a listed company, the regulations are enforced by the Corporations Act.
      • If an investors circumstances change, and they need their money out of the business, there is not usually a ready market that will value and buy that equity. This will cause considerable distraction to the management of the enterprise if it happens.

 

Raising funds is nothing more or less than a marketing project. The entrepreneur has a ‘product’ to sell, the future profitability and potential capital gain from the enterprise. The investor/lender has the funds necessary to crystallise the entrepreneurs dream.

Both lenders and investors require common information, which they use differently. For a lender, it is the reassurance that the loan including principal and interest will be repaid over the life of the loan. Whether that repayment comes from the cash flow of the business, or in the event of failure, sale of the assets against which the loan is secured is largely irrelevant.

In the case of equity, the driving consideration is the potential for capital gain at some point, after the payment of dividends. Both rely on cash flow forecasts from the enterprise.

Agreeing the level of equity that will be exchanged for the investment is a really challenging process. The financial interests of the entrepreneur and the potential investor are directly opposed. The question is how much a point of equity is worth in cash terms. In almost every case, the better prepared party to that negotiation will win. However, it should not be seen as a binary negotiation just about the cash, as there are other variables at work, such as the networks of the potential investor, which as noted above can have significant value.

This process also must place a value on the ideas, and time of the entrepreneur, without which there would be no potential investment.

In short, you need to find a mutually acceptable valuation of forecasts of future cash flow from the ideas and commitment of the entrepreneur, and the value of cash and other factors the potential equity holder brings to the table from which to agree an equity split.

Header cartoon credit: once again Scott Adams and his mate Dilbert understand the dilema