Over promising systems sellers can be the bane of the science researcher’s life, admits Matthew Partridge.
There is an English idiom, ‘call a spade a spade’ which if you’ve not come across it before essentially means call something as it is, be honest, frank and accurate. The phrase is very old and actually comes from the writings of Plutarch before finding its way into English in 1542. So as idioms go, it’s fairly universal, which is helpful for the next bit of this article.
When I buy an item or service, I generally like to know what I’m buying. If I’m shopping for a spade, I generally want to be shown some nice spades, be offered a good price for a spade and then be able to order a spade. Later when a spade shaped parcel arrives, I look forward to unwrapping the reasonably priced spade so I can do all the spade things I want to do.
Why then on more than one occasion have I opened my parcels to discover a fork?
The above tortured idiom is one that I think over in my mind a lot lately when it comes to finding the right tools to do my research job. Particularly in relation to the software packages that are increasingly vital to the work that I do.
For those not up on AI, ‘solving hallucinations’ is the current holy grail of the AI world that Meta, Google etc. have spent literally billions trying to attain
As I mentioned, there are an increasing number of software packages and services that are becoming a fairly core part of the research toolkit. From higher level items such as ELN or LIMS systems to more specific focused tools for domain areas such as [insert third complex acronym here]. A cursory glance round the halls of your nearest lab trade show will give you some idea of how much of an explosion there has been in companies supplying these tools and they are all looking for ways to stand out and be the best.
At one unnamed trade show I was speaking to one unnamed company who promised me that their AI enhanced LIMS system was the best because it was 100% accurate. Two booths down, a second unnamed company said that theirs was equally accurate as they had “solved AI hallucination and so it never happens”.
For those not up on AI, ‘solving hallucinations’ is the current holy grail of the AI world that Meta, Google etc. have spent literally billions trying to attain. Any company that had solved it wouldn’t need to be at a trade show; they would be instantly worth enough money to buy all the trade shows without even triggering a call from their bank… which they would also be able to buy.
This kind of overselling is not limited to AI-infused products. Tough and numerous competition obviously increases pressure on marketing and that leads to some slightly exaggerated claims which in turn leads to a frowning researcher unable to do the thing they were promised as it’s ‘coming soon’… probably.
I’m thrilled with all the digital innovations in research. In many ways this has long been overdue and I couldn’t be happier we’re catching up with other areas of industry. But, for anyone trying to navigate this newly volatile world I would encourage you to get very good at software analysis and sales negotiation (always get a refund clause added). If all else fails, get good at making do with a fork.
More than 60% of life sciences laboratories are exploring or piloting artificial intelligence technologies, but only a small proportion have …
Trust and regulatory uncertainty remain the biggest barriers to artificial intelligence adoption in clinical trials, according to new polling released …
Phesi has published an analysis of more than 600,000 clinical trial protocols, which concludes that fewer than one in three …
© PMGroup Worldwide Ltd 2026
Leave a Reply