Has AI become negative? And if so, is this a problem?

It is a warm, sunny and busy July. I spent today briefly exploring the sense of sin in cross-cultural perspective, reflecting on archetypes, discussing dream cycles, planting three different types of shiso, and planning a workshop for the upcoming DIA meeting in Belgium. In between I wondered about good and bad AI and started to allow myself some happy thoughts about the upcoming summer holidays. My mind is in a happy place full of possibility.

Today’s blog topics:

– Pharma communications and navigating translations
– Content management strategies to keep you sane
– Has AI become more negative? Thinking in the information age
– Unheard harmonies – joy and difference

Any company involved in international business faces the translation challenge at some point. For pharmaceutical companies who interact with patients, health care professionals, patient advocate groups etc. this is a particular challenge because beyond the expense of maintaining documents in many different languages, there is also the challenge of version control.

What to consider as you approach translations?

  • Content type defines translation imperatives: some must be provided in local language, some may have to be provided in local language, some falls into a discretionary category.
  • The market: is your product prescribed by international experts or predominantly by local GPs?
  • Specific Regulations: are there regulations specific to the pharmaceutical industry that address language?
  • General regulations: are there laws that talk about language in the general health care space? For example, the Finish health care act section 6 which addresses patient care and patient rights concerning the language of care provision.
  • Cultural context: How is content provision in English perceived? Does the entire document need to be in local language or is a mixed approach, cover letter in local language, standard response in English acceptable?
  • Diversity and Language: the written and spoken word, vocabulary and nuance varies between countries and regions. This is true for English, French, Spanish, Portuguese, German to mention some of the more widely spoken languages.
  • Translation approach: case by case? Or all documents? Or only frequently used documents?
  • Currency: how do you ensure all child documents are current?

Key take-away: Approach document translation with caution and assess all variables before committing to an approach.

Pharma content management strategies to keep you sane

Historically departments generated content independently of one another. However, this approach leads to both inconsistencies and an incredible workload full of “busy work”. As teams do more with less and share more and more content across different channels, reducing reduplication of effort will help you stay sane.

Some suggestions:

  • Identify
    • the type of content you use now
    • the type of content you will need in the future
    • the format of the content and channels
    • which teams require which type of content
    • which content can be used across teams
    • required core content and source documents
  • Agree
    • harmonised formats/templates – with without colours and branding etc. as appropriate
    • a version control approach and naming convention
    • the content lifecycle (review, approval, sign-off, use, retirement and archiving process)
    • access policy
  • Implement
    • a master data management approach (source data is stored in a single location and used from there)
    • identified data owner(s)
    • a global data governance approach
  • Consider
    • a concatenation approach to content/document generation
    • a global approach to content generation
  • Involve
    • Key stakeholders from business and IT early
  • Research
    • Systems that will scale and easily integrate with your current system architecture

Key take-away: Many teams use very similar content, having a harmonised approach will accelerate and facilitate your work. The use of AI is further down the road than you might imagine.

Has AI become more negative? Thinking in the information age

Two weeks ago, I was at a conference. A presenter was waxing lyrical about the powers of ChatGPT. He shared a graph by Rita McGrath, professor at Colombia Business School that was published in the Harvard Business Review in an article entitled “the pace of technology is speeding up”. The original graph title was: “Consumption spreads faster today” and the text read “innovations introduced more recently are being adopted more quickly” You can find the original article and graph here. The title the presenter used on his slide was “good news, humans get faster” my first impulse is to think “yes, faster is better” then I looked at the graph again and asked myself “does this graph show me that humans are faster? And is it good news? If so, why?” I reviewed the initial data

The presenter also shared data from a paper with the title “Did AI get more negative recently”. In brief, the paper assesses the stance of authors performing research in the Natural Language Processing and Machine Learning Fields. He shared the information that negative publications are cited more frequently than positive publications, and his opinion that this could bias the general public against AI and that this is a worrying development, because in his estimation AI is good.

Some thoughts from me 1) a negative stance towards a certain topic doesn’t mean the author is wrong 2) the publication focuses on scientific literature, which is not in the public domain, hence the public is unlikely to be exposed it 3) the fact that negative publications in the data set that was analyzed are cited more frequently than positive publications is neither good nor bad, just interesting.

In the cited publication the authors point out that there are many reasons why negative publications in this field receive more citations, they also proffer a hypothesis that the increase in critical publications in the Machine Learning and Natural Language Processing field in recent years may be a natural response to the hype and a peak of positivity in 2010 when “people started to challenge the validity of some of the claims, when issues of adversarial robustness and reproducibility became apparent and people began to question evaluation frameworks” You can find the peer reviewed article on the Royal Society Open Science website.

The reason I am sharing the above is that with the internet graphs are easy to find, and copy, good quotes are “likeable” and cursed with little patience and less time it is easy to be misled by a good story. Humans like to take data and make it fit their belief system. And this is then shared over and over. Every day I read stories that sound great but that are not backed up by the references that are provided.

Key take-aways: 1) Think for yourself, check the source documents, and reach your own conclusions before sharing, liking, or believing. 2) If it is all black or all white it is unlikely to be true.

Harmonies unheard – joy and difference

I met a woman on the street the other day. We chatted for a while. Her speech pattern was unusual. She asked me to enunciate clearly, to speak slowly and to face her as I spoke. Then she told me that she had been born deaf.

She told me how important music is to her, that she regularly goes to classical concerts and that she especially loves opera. She described the way she experiences music; the vibrations and the unbridled joy music elicits in her.

As I cycled away, I had a big smile on my face, her pleasure was infectious. I was reminded that the world is a big place, that humans experience the world in a myriad way and instead of putting myself in someone else’s shoes it’s much nicer to let them tell me about their world.

Key take-away: difference is all around us and the best way to understand another person is to listen to them with interest and an open mind and heart.

I hope my blog provides you with useful insights and I look forward to hearing your thoughts. If you have a challenging project in the realm of medical, digital, systems, analytics, channels etc. or you are navigating a team or personal challenge contact me for an informal chat and to see how I can help.

Best wishes

Isabelle C. Widmer

Photo Credit: Markus Winkler @unsplash