Changes at DAM News
We have been making a few changes here at DAM News which I wanted to make our readers aware of. Firstly, my co-contributor, Russell Charles McVeigh is now the editor and will assume responsibility for all day-to-day operations.
The weekly news round-up, events and job listings will continue as usual. Further, we are still accepting articles from external contributors (subject to them meeting our editorial guidelines). Currently, I remain the proprietor and publisher of DAM News – although that too may change. I am also winding down my DAM consulting business, Daydream and no longer getting involved with large-scale Enterprise DAM and Digital Asset Supply Chain implementation projects.
So what are the reasons for all this? The first is health-related. With impeccable timing, shortly before the Covid pandemic got going in early 2020, I was diagnosed with a form of Chronic Kidney Disease called IgA Nephropathy. The doctors refer to my condition as ‘a common rare disease’ (an oxymoron which the medical profession seem uniquely capable of devising). What the outcome will be, whether I will need dialysis or transplants etc. at some point is impossible to predict. Similarly, no one knows why I have acquired it in the first place (although I can count myself lucky it was not until later in life, unlike those who get it in childhood). There is a good possibility that drugs might become available to treat the condition, but no one knows when, nor if they will actually work.
Faced with this existential threat which might potentially short change me of my three-score and ten, some difficult decisions have needed to be made. I have worked in the DAM field for 28 years now and my considered opinion is that it hasn’t lived up to its promise. The opportunity offered by the increased demand for DAM which has been in-play for at least 20 years or so has essentially been squandered by the current cohort of actors in the DAM industry. I would, unfortunately, have to consider myself among their number, despite my best efforts.
This market has been a demonstration of what happens when you get a niche tech sector suffering from a poverty of new ideas, combined with a febrile concoction of group-think, cliques, cabals, under-investment (or poor capital allocation decisions when money was made available) combined with an ever-present fear that the whole artifice could come crashing down at any moment.
To borrow one my Father’s colourful expressions, both the DAM industry and its self-appointed leadership are, in my opinion, NFG. I now want to go and do something else more productive and interesting with whatever time I have remaining.
Before I do so, however, I want to take a final opportunity to analyse how I see things shaping up for the DAM market, both good and bad.
Unquestionably, the driving force behind the growth of DAM in the last 20-30 years has been the exponential growth of archived digital assets. The recent introduction of Generative AI (GenAI), however, potentially disrupts this factor. In the future, there may not be the same reliance on archived material if it is possible to generate what you require from a text prompt instead. It is true that GenAI won’t be a suitable source for all content. There are still significant numbers of digital assets which are likely to be impossible to generate via algorithms alone, but even so, this technology will undeniably have an impact and it will be a progressively more influential one.
The option to search archives and sift through results will be far less appealing than the ability to generate what (in supply chain management terms) might be called ‘Just in Time’ content. This is what GenAI promises and why it poses a threat to DAM. With that said, there are mitigating factors and they are being underestimated, at present.
There is currently a huge amount of irrational over-exuberance when it comes to AI. When I discussed the GenAI subject last year, I postulated that there would be greater demand for photo re-touching skills. The crafting of prompts to drive AI will become a key skill and this is being recognised already. In many cases, however, repeatedly trying to devise inventive prompts to coax out what you require will take longer and might be less effective than simply finding someone with the necessary production skills to manually make the adjustments instead.
The same principles will apply whatever the medium, whether it is visual imagery, narrative text, music, computer code, 3D graphics and so on. This is the Achilles Heel of GenAI and I think it will remain one for a long time. I mentioned this effect on a few Linked discussions I have got involved in recently, an approximate analogy would be something like the children’s story about the three wishes.
If GenAI repeats previous AI patterns, the initial excitement will diminish when it gets applied to more complex real-world tasks. This is when flaws emerge and compromises have to be made. These will necessitate time being spent on manual adjustments and the benefits from GenAI start to become increasingly marginal.
The ‘intelligence’ in GenAI is largely derived from data which has been supplied by human beings, often without their permission (which is another legal risk not being fully appreciated right now). Further, the algorithms are quite significantly inefficient at present – and the problem appears to be getting worse, not better. This post by Jennifer Prendki illustrates this point:
“… when the model doesn’t deliver, ML scientists like to make it more complicated. That’s because the way we do AI right now is not “modelling” at all – it’s just information encoding. Forget if we don’t understand how, or even why they work. Forget we have no way to identify biases. As Timnit Gebru would say, we’re building the perfect stochastic parrots… Call me an idealist, but I think AI researchers have a lot to learn from physicists and scientists in general. As far as I am concerned, I believe we still have a lot more work to do for AI to become truly intelligent.” [Read More]
The computational cost (processing power) required to keep feeding these inefficient algorithms is likely to become an increasing problem. Earlier this year, I had a discussion on LinkedIn about this quote by Sundar Pichai (CEO of Google): “Today, the scale of the largest AI computations is doubling every six months, far outpacing Moore’s Law”. I don’t believe in Moore’s law and see it as (at best) an observation made at a point in time rather than a ‘law’. Nonetheless, you don’t need to be an expert in AI technology to see the potential problem on the horizon.
This is an ‘immovable object meets an unstoppable force’ type of scenario. AI might represent the unstoppable force, but the immovable object is finite processing power being insufficient to support the ambitions of AI/ML algorithms which prioritise visible results over efficient design.
What we are seeing now might be a temporary high point in AI and there could well be a period of sideways movement, or even a decline. I took the recently publicised open letter signed by a number of dignitaries to pause AI development as a signal that AI may well have reached a temporary plateau. These are indicators of people buying the hype wholesale and not applying critical judgement, nor carefully analysing the facts.
Temporary roadblocks in the progress of AI are potentially positive for low-innovation tech sectors like Digital Asset Management since it means the decline in the use of archived content could be slower and far more gradual than it first appears. This gives the DAM market longer to get its house in-order when it comes to genuine innovation (with or without AI).
Whether DAM software developers, systems architects and product managers will be able to re-imagine a credible reason for their products to continue to exist outside the conventional archived content remit where they have historically been based remains to be seen. Simply plugging into other technologies will not be sufficient, however. Ultimately, you can’t run a successful tech business and outsource innovation to third parties who are indifferent to your success or failure. I believe that how the industry handles this transitional period will define whether or not it survives.
So where does that leave us with DAM News? I’ve been the propreitor of this title since 2014 (and contributing since 2012). It provided a way for me to stay in-touch with multiple industry contacts, a platform to present my ideas and also generated a steady stream of consulting sales leads. In the early stages, the latter partly subsidised the cost of maintaining DAM News and building resources like the glossary, vendor directory, and open specification etc. The premium reports, subscriptions and advertising revenue has increased in the interim period, but it has never been fully exploited because there was not really a need to do so.
For tool vendors, consultants or others who operate in the DAM sector it could present an opportunity to gain access to a ready-made audience of thousands of people who are closely involved with the industry (on both the buy and sell sides). There is also a substantial body of authoritative content dating back to 2009. As far as I am aware, DAM News is the longest running DAM-specific trade news publication.
I am open to sale and/or sponsorship of DAM News by anyone who thinks it might offer them some value. My co-contributor will remain indefinitely to manage the editorial operations for as long as is required. If you are interested, contact me: ralph dot windsor at daydream.co.uk and we can talk through the details.
In terms of what I plan to do with myself hereafter, writing (albeit probably less about DAM) will likely be something I continue with. As mentioned, I am not open to helping anyone out with their DAM or Digital Asset Supply Chain implementations, however, I am still interested in talking to vendors. In particular, tool vendors and smaller firms; or to be more accurate, vendors who don’t feel the need to pretend to be much bigger than they really are, since in reality none of them are particularly large.
I have had some interesting discussions recently and a few more open-minded firms have understood the points I have been making for a number of years about value chains and how to more effectively utilise AI for DAM. It would be good to continue these and help them to develop something both useful and innovative. It’s been far too long for very little to happen in this market.
Since this may well be my last ever post for DAM News, I would like to take this opportunity to thank everyone who has contributed to my knowledge about this subject over the last 28 years. Although I have often presented a somewhat jaded and cynical perspective on the industry, it is because I genuinely wanted to see it realise its full potential and not settle for the kind of complacent, self-congratulatory navel-gazing which it is (sadly) well known for. If there are regrets, it is that I did not pursue this objective far more aggressively than I have done. This is now a task for others to undertake, however and I wish them every success in that endeavour.Share this Article:
Ralph, I have always valued your posts and found your pessimistic and charmingly curmudgeonly take the most reliable perspective on DAM that could be readily found. I was so sorry to hear about your diagnosis, and will keep in my prayers, hoping that all turns out for the best and you get your threescore and ten—and many, many more. I will definitely be contacting you about the things that remain of interest to you, if Pica9 is the sort of firm that would fall into that category. But no matter what, I want you to know that I have always found your work clarifying and even inspiring, and I’ll miss your perspective as this category of ours stumbles along forward
Ralph, I too have found every interaction I had with you interesting, informative, and enriching. I send the very best thoughts to you and hope for healing. May this forced change in life bring you unexpected opportunities to do things that fill you with joy.
Thank you Ralph – been a loyal reader pretty much since I switched career paths towards DAM in October of 2014 – somewhere around there – added the RSS to my Feedly and just kept finding new nuggets of insight to keep me coming back. I’ve learned a lot from DAM News and wanted to let you know your efforts are appreciated.
Very interesting comments about GenAI!
Do I think that just in time content generation will become a future state? YES and it will have it’s own place.
Do I think GenAI will impact DAM? Not really. Why … because of Authenticity. There is both a need and a desire for people to experience authentic content.
What will impact DAM and is all to do with AI is the ability to easily locate content for a user using some simple phrases.
Hubspot is our CRM vendor. Just recently we had our monthly update briefing and YES Hubspot now have ChatSpot, a rebranded form of ChatGPT. Of many things that ChatGPT could do for a CRM user the one that Hubspot are looking at in depth is to remove the structuring of the interface to find all the various bits n pieces that we/I use …. and replace that with what will most likely become voice control. “Hubspot display my dashboard for marketing!” Gone will be the login, go to dashboards, select existing template or create new. A few spoken words will get me what I want.
OK sounds great but what has that to do with DAM Ricky!
“DAM show me all the most recent photos showing regrowth in the forests of southern NSW after the fires of 2019-2020”
Now if we wanted to provide the right content, in the right format, for the right person who asked that question …. using a DAM …… hmmmmm I’m sure many of you are ticking over with table schemas, metadata automation and more.
How about AI recognition of a big bucket of photos?
With a voice command AI driven search and retrieve front end?
How would any of our current DAMs, which take XXX hours and YYY dollars to build, compare to that?
I wish you all the best Ralph.
Kevin Groome, thanks for your kind words, I should get badges made with ‘charmingly curmudgeonly’ printed on them! We’ve had some great conversations about DAM in the past, please feel free to get in touch. Many thanks for your ‘Confessions of a DAM Vendor’ article article. For those who haven’t read it, check it out here: https://digitalassetmanagementnews.org/features/confessions-of-a-dam-vendor/
Magan, I think it’s fair to say you are one the definitive authentic voices in DAM with decades of proven expertise in this field and I really enjoyed working with you. I still refer people to this great article you wrote in 2018:
Kevin Yezbick, thanks so much for your kind words. It’s gratifying to know that people got something useful out of this stuff and demonstrates that there is still a need for a trade publication about DAM with some serious discussion of the topics surrounding it.
Ricky, good to hear from you again!
Over your points, I think there will still be a demand for archived content that can’t be produced by GenAI. As I mention in the article, I suspect a lot of this stuff will still need someone to re-touch or amend it – which will create new assets that still have to be archived somewhere.
With that said, when GenAI becomes widely available, the user logic is going to be something like “I need x content, I will type some keywords into a text prompt to see what I get.”
If it’s a very specific item, they will probably go for an archived asset, assuming a suitable one can be found from whatever keywords they enter. If it’s a generic ‘stock photo’ type image, they’ll probably be satisfied with GenAI. The latter might also have the advantage that it will be unique (lots of DAM users have reported to me in the past that they ended up using the same stuff all the time). Returning to the first case, if they can’t find a suitable asset because it hasn’t been catalogued effectively with appropriate metadata, then they are more likely to use a generated one instead.
The reason why GenAI may erode DAM is that users may not always require authentic assets and if it’s a case of having to waste hours of time trying to track down a ‘real’ asset, they might decide it isn’t worthwhile, especially if they are very busy. A lot depends on what the task at-hand is and how critical it is that they get an archived item. Keep in mind also that the GenAI tech is probably going to get more sophisticated and skilful at rendering content which is usable for many more scenarios.
Much also depends on how DAM vendors treat the requirements of their users. Too many assume that what they are creating is like an in-house version of Shutterstock or Getty etc. Most Enterprise DAM users aren’t actually looking for a stock photo replacement. Saving money on not buying commercial stock is an issue, but less of one than finding the exact image they need of one of their products, a specific employee or some event that took place at a given point in time. Currently, too many DAM vendors don’t really get the needs of their users, they have to be told by them what to do and how their solutions should work in order to meet their unique needs. If vendors persist with this approach, they will lose out to GenAI (or to be more precise, something basic like Dropbox which can also interface with GenAI).
Over the AI classification of assets, again, generic cataloguing simply won’t cut it. The issue is context. See this article: https://www.cmswire.com/digital-asset-management/what-it-will-take-for-artificial-intelligence-to-become-useful-for-dam/
From what I’ve seen of things like ChatGPT, they have some ability to contextualise, but it’s quite basic and they still need to be directed, sometimes with multiple prompts. The issue where ChatGPT returns different results each time is (in equal parts) a blessing and a curse. It’s a black box and you can’t really direct it to behave in a particular way, which is a problem for a specialised use-case like Enterprise DAM.
Many of the DAM vendors don’t provide metadata cataloguing facilities which are sophisticated enough for the needs of their users. There is an implicit assumption by some that all this will get solved by AI. The reason behind this is that it’s hard work developing advanced metadata management tools which can be adapted to the needs of many different users. If they don’t, however, in the GenAI era, they effectively remove what might be one of their few remaining USPs.
For all these reasons, providing useful DAM software is going to get harder. Simply plugging into ChatGPT, GenAI tools, LLMs etc. will not be sufficient. If vendors find this kind of quite specialised thinking difficult to deal with, or unappealing, their best option would be to recruit many partners in different vertical industries or who specialise in a key area. This is really where the value in DAM will lie in the future, it’s the ability to efficiently specialise toolsets to the unique needs of individual clients. I do believe there will still be demand for that from DAM users.