I note the comments if they are in possible reference to my own responses so far. However, I am not an academic. I suspect many front line social workers (whether in case holding practice or not) are also not academics. Other contributers sharing their knowledge in AI to this article and with interest to assist the profession, may also not be academics either. We all do however make necessary study research reflections.

The issues of AI in social work is here as you rightly and correctly say. They are also as you say, impacting day to day social work practice. Getting social workers away from computers what AI can do in seconds, as you say. It seems AI also deeply affects the human approach towards service users engagement. This includes in wider workforce, CPD learning, education and application of professional standards. Ironically, I have wondered why the Social Work England sit within the UK government Department for Education (DfE) and not strangely, with the Department for Health and Social Care (DHSC). May be because of Education funding and services relating to Children & Families and Adults dichtomy.

Your comments about costs and predictive conditions for social workers with impact of AI roll out seems also shared elsewhere and what you suspect will really happen:

https://www.mysocialworknews.com/article/ai-can-help-write-casenotes-quicker-but-can-it-do-a-home-visit

The point remains, and relates to the Fitness to Practice processes. It connects also with the application use of requested submitted diversity data. They have shown to date an official outcome by the Social Work England, pointing towards discrimination practices. The outcome is linked deeply and possibly ‘harmonised’ in all the stages and across affecting Children – Adult workforce. What those affected still do not have is the official acceptance nor to the role of AI in such processes (if there can indeed be any contributions), to the findings.

Social work England need to focus on meeting their own professional standards and gaining a basic comprehension of the profession before even thinking about AI.

With context to the Fitness to Practice process and experiences, I am thinking perhaps ‘old school’ from a social science (not computer science) perspective. These are to the potental issues and learnings arising from the discriminatory findings.

In relation to AI, utilisation of ethnic diversity data and the communities showing most affected, it does not seems because of, nor interpretated as ‘failure’, in AI usage, elements or aspects archictecture. Instead, perhaps more in the opposite of a ‘success’ statement being disturbingly played out and reported – as achieved. In particularly, the deeper and interpretative use of protected and private data, whether by ‘big data’ machines or in this matter sensitively needed by regulatory professional bodies, requesting urgently ethnicity & diversity data to inform ‘gaps’ for best organisational outcomes within un/consciously bias frameworks.

The findings from Professional Standards Authority, and other regulatory landscape bodies do not seem to be showing ‘failures’ or decline in their reporting of discrimination if and when asked. This is within the health and care profession and perhaps other professions. I am not aware of reported decline trajectories in experiences, but instead the opposite and or status quo, observed normalised into daily operations, unless proactive measures in place.

I am perhaps of old school as said. I am mindful of and will continue ponder respectfully to the important comment of the above author –

“….Unless AI is confined to highly standardised, deeply harmonised and routine functions it will inevitably fail…”

‘The types of AI being used across health and social care in England and their application in social work practice, including the risks of bias and discrimination.’

This Artificail Intelligence (AI) aspect including for sections of the Black workforce community, is and remains point of keen interest. Not least with reference to Fitness to Practice processes. Unfortunately, to best of knowledge there has still been no valid offered explanation by the SWE to the facts of its findings. This is despite borne out by its repeated, consistent and rigourous data crunching analysis. Having responded to the regulators’s important request to submit ethnicity data, the findings have shown systematic demographic discrimination repeated impacting particular groups.

In most simplistic term, it seems the new announcement of this AI probe – is a case of the ‘tail wagging the dog’. Its back to front, delayed and seems not a proactive initiative on the part of SWE. This is when you consider the historical work already reported to and commissioned by its accountable body the Public Standards Authority (PSA) 5 years ago. This is with reference to the ethical aspects, data, and advance propose usage of AI going back even further in 2019.

The links below are seperate with context to AI in health care profession and impact experiences of discrimination – where in the latter, the value of reported lived experiences also count as valid data informing strategy and boardroom decision in health and care professions.

Please see;

https://www.professionalstandards.org.uk/news-and-updates/news/artificial-intelligence-what-it-and-what-impact-will-it-have-professional

and associated:

https://www.professionalstandards.org.uk/news-and-updates/news/racial-discrimination-healthcare-regulation-and-oversight-importance.

Additionally, I sense this article 4.2.25 is also a way of SWE preparing to advance the reasons for the discrimination (in relation to Fitness to practice processes). That is because of or a contribution of AI. While not pre-empting, I have commented elsewhere on the moral and ethical implications when and if we rely on the position of AI for accountability. That is, you cannot hold ‘machines’ or ‘robots’ no matter how sophisticated, developed and advance or complex, accountable. Not when attitudes, behaviours and decisions are made by people.

The Zachmans Inventory and the Carnegie Mellon Model both accepted measures used by TOGAF (the open group architecture framework) and adopted by, then, UK Government in 2010, to assess organisational value and competency (the resource view of the firm), showed AI failed!

The level of accuracy driving down standards by reducing social complexity to basic level computational ciphers based around normative assumptions about all behaviour having the same properties and qualities, and measuring how consistent the information gathered is with and from these norms ie all water is wet type judgements. It’s the idea of completeness assumed by Booles when arriving at the yes/no (one’s and zero’s) in lie detection. The specificalities of the design criteria in use is surveillance the rathermore nuanced elicitation of any espoused values simply incompatible with the design briefing.

Sha Xin Wei warned against such catastrophic uses of data in his seminar ‘Navigating Indeterminacy’ presented to the Australian government a few years ago. And, decades ago Eric Trist and Howard Permutter writing in 1986 warned against the same, see ‘Societies in Transition’ Journal of Human Relations 1986. As well as the huge and costly failures in IT procurement across the NHS and LocGov since Interoperability was first tested.

The very first IT kit bought by Westminster Council for use in Children’s Services Social Work, called UCare back in 1983 (I knew the peeps developing the kit) failed, even though the IT provider has changed, it still doesn’t work!

Unless AI is confined to highly standardised, deeply harmonised and routine functions it will inevitably fail.

Any coder will tell you this. Speak to the DnD gamers who have been doing this for decades They know that the DM makes or breaks the game and the players.

Are Social Workers, as the so called navigators, the new Dungeon Masters of The Poor Laws Wars; Or What?

For a real time and real world take on this check out Dunder Moose a US podcast covering the ethical responsibilities of a DM towards the players.

Crazy, crazy 🤪