You are on page 1of 10

Foresight

State of horizon scanning in 2018


Examples in occupational safety and health

The foresight exercises led by INRS since 2015 have structured horizon scanning devoted to the
development of production modes, with a view to detect the appearance of early “signals” of
transformations that could affect occupational risk prevention. This article presents highlights in the
world of work, identified in 2018 and selected either because they confirm certain foresight
assumptions, or because new elements have appeared, requiring future work situations to be
apprehended in advance.

This second edition of the state of INRS’s horizon scanning builds on the first edition
published in 2017 (see For more information) on elements identified in the press or in grey
literature. The goal is to select concrete examples and compare them to certain assumptions
formulated during three foresight exercisesi conducted by INRS, regardless of whether or not
the examples are in line with those assumptions. The foresight exercises were all devoted to
the development of work conditions and occupational risks, considered from the point of
view of work organisation and changes related to the digitalisation of activities:

 use of industrial exoskeletons by 2030 in France;


 healthy working for 2040 (production methods and modes in France for 2040);
 platformisation 2027.
Matches or contradictions can be purely contextual. They illustrate the evolution of topics
considered as emerging topics at the time of the foresight exercise. Continuing this horizon
scanning exercise over several years should improve the understanding of changes in
progress.
A limited number of references is provided in this document: they have been selected
because they illustrate certain trends identified during horizon scanning performed in 2018.
Some other references, which are not related to these automation issues, but which could
be precursors to the development of occupational health and safety risks, are presented at
the end of the document.

An example of excessive use of robots, generating industrial accidents and an increase in the
occurrence of occupational risks
Elon Musk, with its multitude of industrial activities, was all over the news in 2018. The
financial and technical difficulties of its company Tesla, came under heavy scrutiny: the
difficult rise in the production cost of the model 3 sedan brought the company close to
insolvency. The project was particularly ambitious, aimed at the wide-scale production of an
electric car at a “reduced” cost ($35,000ii), to give a broad public access to what Elon Musk
describes as the car of the future, all the while making a profit iii. Therefore, the company was
not interested, like other carmakers, in having an electric car in its catalogue, but in making
it the central element of its entire industrial strategy.
The difference is not only in the car’s technology, but also in its production organisation.
While the manufacturing lines of other carmakers are designed to function with workers and
robots for a portion of the task (most strenuous operations, most reproducible, easiest to
automate) Tesla’s challenge was to design two entirely automated assembly lines.
While some great technical success was seeniv, the functioning of certain robots was a bit
random, and even deficient for long periods of time. The technical problems were such that
the company had to urgently set up a third assembly line outside the factory’s production
buildings, under a temporary structure (an aluminium frame tent), with human workers and
robots (most of which had been used in trials or were taken from other production lines) v.
These setbacks had consequences on working conditionsvi. While Tesla’s occupational
accident rate was roughly 30% higher than that of its American counterparts, it dropped to
normal levels as from 2017. But strikingly, while the percentage of serious accidents at
Tesla’s counterparts was around 60% compared to the total number of accidents, at Tesla, it
reached 90%, providing grounds for investigations by the labour inspectorate of the State of
California where the factories are located. In 2016, the rate of occupational accidents
requiring sick leave or job restrictions at Tesla was 83% higher than in the rest of the
automobile industry. In addition, cases of exposure to allergenic or reprotoxic chemicals
were cited.
Disagreements emerged concerning the qualification of a certain number of accidents and
health problems. Although these cases appeared after accidents occurring at the site during
work, or presented clues that suggested they might be related to work conditions and
recorded as such by local management, they were qualified as “personal medical cases” and
their occupational nature was denied by the company.
On the other hand, several testimonies incriminated the production organisation system
where machines had become key, to the detriment of workers, and where multiple robot
malfunctions in the production lines had consequences on the occupational accident rate vii.
Assessing the functioning of the factory manufacturing the model 3, Elon Musk tweeted
“Excessive automation at Tesla was a mistake. To be precise, my mistake. Humans are
underrated”viii.
Robotisation aimed at supporting workers and facilitating their tasks
Materialising the announcements of a collaboration with Eksovest made in 2017 ix, Ford
announced a procurement plan for 15 factories throughout the world: to equip workers with
exoskeletons aimed at facilitating very physically demanding tasks, which require operators
to work with their arms raised above their heads. These were passive exoskeletons,
therefore not powered with motors, that simply provide assistance to reduce the constraints
on the workers’ arms. The mechanism is a vest equipped with arm-support devices. A short
film shows how it worksx. The carmaker highlights the drop in constraints and fatigue, and
not an increase in production rates. In another similar example of use, the occupational
health and pension fund of the Auvergne region in France accompanied process
transformations led by a poultry slaughter and processing plant, in which uncomfortable
working postures, in particular major bending of the torso and the upper limbs and carrying
loads above the head, were performed throughout the entire work shift. The use of
automatic basket stackers, enabling work to be performed at constant height, and a cobot
handler for moving stacks reduced these constraints. This workstation transformation can be
viewed in a film made by the equipment manufacturerxi. The main elements of the change
are shown in Figure/Photo 1xii.
The transformation made the workstations more inclusive (accessible to women and men
regardless of their physiological features). It was done without any changes to staff and
without increasing rates. After a few months of operation, the employer however noted an
increase in productivity. This seems to illustrate a major potential for the development of
robotised technology in medium-sized companies, of an average technological level: cobot
solutions, relatively simple and financially accessible have begun to emerge, reducing
physical work strain.
In terms of automated handling, many changes are to come. Man’s intelligence and
dexterity currently give him a sure advantage over machines, for tasks that require dexterity
and intuitive capability. In cases where machines reach par with humans, it is often at a
prohibitive cost. However, recent developments show that this solution could evolve in the
medium term. An autonomous hand (the spinner) which reproduces human gripping ability
through an opposable thumb, was developed. The robot can also recognise the shape of
objects and adapt the position of its fingers. In the same vein, “the picker” can distinguish
certain objects based on their shape. By referring to an initial database of 10,000 shapes of
objects, it can determine the most suitable technique for picking any object, including if it is
not in the initial data. Connected to a neural network algorithm, it can learn from its
mistakes and its achievements and develop its capacity, depending on the circumstances in
which it is used. Other examples of these technical evolutions have been published xiii.

A study devoted to the Italian industry of the future


The product of a study conducted in several large industrial establishments in the north of
Italy, a book was published in 2018xiv. It has many objectives: for actors to express their
views on the current production transformation, both from the technical and cultural point
of view; unveil the major ongoing trends and those that are expected to shape the future.
Not unsurprisingly, leaders’ positions support the agility factor: flexibility of production and
working hours, the need to continuously adapt to clients’ demands and the need for
operators to be versatile. Concretely, the new possibilities offered by automation of tasks
(both in industry and service) also challenge the model of “refocusing on core business”, in
vogue these last few decades. Outsourced activities can now be insourced, most often
because the nature of the work and the technology used have changed.
The authors stress the fact that the term “Industry 4.0” does not cover a single model: the
paths followed currently by the various companies concerned are very different. They are
not perceived by the players as an industrial revolution, but rather a smart shift in industrial
production: many activities still depend on humans and the role of robots is still limited to
execution (low level of communication, passive “intelligence”). They contribute in particular
to increasing workers’ involvement and rationalising production. Neither has there been a
revolution in organisation models. In reality, it is more a matter of “flexible automation” or
“toyotism” or “a refined development of post-Fordism”, than a new organisation model, or
as summed up by the slogan: “Industry 4.0 is an augmented post-Fordism”.
How can humans keep the upper hand regarding the ethical matters raised by algorithms
and artificial intelligence (AI): the position of the French data protection authorityxv
In a report by the French data protection authority (CNIL) published at the end of 2017
which summarises its public debate on the ethical challenges of algorithms and artificial
intelligence, the authority formulates six recommendations:

 foster education of all players involved in algorithmic systems (designers,


professionals, citizens): this measure is aimed at raising the awareness of all players
(including public players) of the need for balanced and “symmetrical” use of
algorithms;
 make algorithmic systems comprehensible by strengthening existing rights and by
rethinking mediation with users: service platform experience (food home delivery for
example) shows that it is impossible for workers to have access to a person that can
actually correct malfunctions immediately;
 improve the design of algorithmic systems at the service of freedom, to prevent the
“black box” effect: this is to make humans accountable and increase their capacity to
make informed decisions, instead of taking away their capacity to make choices to
benefit the machine;
 create a national platform to audit algorithms: this is aimed at ensuring their
compliance with the law, which more involves checking the results of their use and
not really examining source codes;
 increase incentives for research on ethical AI and launch a major participative
national cause around a general-interest research project;
 Strengthen ethics in companies: creating ethics committees, spreading good
practices in each sector or the revising the codes of ethics proposed by CNIL could be
perfectly suitable in an approach to regulate the use of AI in the workplace.
Use of algorithms for collaborative remote work
The previous state of horizon scanning (2017; see For more information) focused quite a bit
on “Foundry”, an algorithm aimed at facilitating work in project mode, through online
recruitment of collaborators and assistance with proofing and organisation of contributions.
At that time, we highlighted the possible consequences in terms of occupational risks,
related to the fragmentation of the tasks performed, the absence of visibility into the
purpose of the work done, the undermining of the work community, the impossibility of
disconnecting from work, etc.
In collaboration with Microsoft, Publicis launched its own collaborative platform, based on
artificial intelligence: Marcelxvi. The goal was to provide the 80,000 collaborators in the
group, spread across 130 countries, with a tool to develop their capacity to work together
daily, across borders. The artificial intelligence, based on which Marcel was built, makes it
possible to form teams bringing together all the skills necessary to develop a project, to
draw on experience acquired in similar topics, and to bypass mundane bibliographical tasks
by quickly summarising the useful resource availablexvii. As with Foundry, Marcel is a tool
that has a major influence on work and the conditions in which work is performed. The
images in the film (see below) show the extent to which the mechanism can be intrusive and
complicates the separation between professional and personal life. This is exacerbated by
the very international nature of collaborations within the Publicis group. While it is likely
that a form of regulation will be decided quickly within the group, there are doubts as to the
ability of outside collaborators (contractors, freelancers), who can be involved in the
mechanism, to establish limits faced with the demands that may be placed on them. The
company states that the development of Marcel will represent a considerable investment
cost over the upcoming years, with a major impact on work organisation. The use that can
be made, in terms of human resource management, of such a tool, capable of tracking all
interactions, is problematic, because it could make it possible to identify the employees that
contribute the most (and the least) based on data that do not necessarily represent the
reality of the work performed.

Hacking
After a 37% increase in 2017, cyberattacks targeting industrial monitoring systems again
increased 40% in the first half alone of 2018xviii. It has become a major issue, linked with
different motives:
 cyber-extortion, making companies have to pay for their systems to be unlocked;
 theft of data, such as operating modes, patents, customer sheets, technical
specifications, etc.
There is a third motive: to destroy facilities, or cause human loss among the company staff.
The most spectacular example is the cyberattack on a Saudi oil refinery, which was aimed
not only at paralysing the plant but also causing it to explodexix.
The virus attack on the plant (which affected voltage and pressure regulation instruments)
has not been elucidated (or at least not made public), but it appears certain that the attack is
not due to the instruments being hacked before they were installed in the plant, nor to an
inside job. Even more disturbing, the regulation instruments targeted are present in 18,000
industrial facilities throughout the world. The failure of the attack was not due to the
facility’s antivirus protection, but to the presence of malware in the virus programme which
itself had been contaminated previously.
The initial elements of the investigation point to an operation ordered by a neighbouring
State (Iran) and led by groups that it controls. Additional elements suggest that other
parastatal (Russian) players also had a role in the affairxx.
Lastly, it seems that a portion of the viruses used in the attack against the Saudi refinery was
used before in attacks on Iranian uranium enrichment facilities xxi (led by Israel and the USA).
The magnitude of the problem has led the USA to consider these matters more seriously. It
appears that they are changing their stance from an essentially defensive one to a more
systemic approach, aimed not only at improving defence systems, but also at waging war
against the different groups identified in certain countries that lend their support, or conduct
these cybercriminal operations themselvesxxii,xxiii.
Past experience has shown that low-level cybercriminals quickly adopt sophisticated
malware: this was observed with operations aimed at relatively small companies (see For
more information).
On another note, the last few years have seen a democratisation of equipment for
performing different DNA operations, with them becoming accessible to simple individuals
or available in biotech labs (similar to fab labs, making it possible to go from the design
phase to the prototyping phase)xxiv. Some of them have reconstituted different variants of
smallpox using DNA freely available for sale, and published the method. Today, concerns are
emerging about hackers’ ability to hack biotech company data to make changes to products
manufactured, which could be potentially hazardous for the workers in these companies,
and even for public health.

Continued social insecurity of platform workers


To say the least, the year 2018 did not see any stabilisation in the professional context of
platform workers. The discontinuation of the food delivery platform Foodora in France in
August, confirmed that the economic model of these new players is not stable. This platform
had only 50 employees in France, but provided work to more than 1,500 independent
delivery persons in eight cities.
In legal terms, many concerns also persist: while questions about work relationships and
social protection of platform workers have become essential, the answers have not yet been
stabilised. Having observed the difficulties in applying Article 60 of the law on the social
rights of platform workers, deputy Aurélien Taché proposed an amendment to the law on
“the freedom to choose one’s future career”, providing for “a responsible framework for
relationships between platforms (…) and workers that use them” xxv. This text opened up the
possibility for platforms to establish social accountability charters, explicitly dealing with
occupational risks and their prevention, having to specify in particular, “the prevention
measures for occupational risks to which workers may be exposed given their activity and
the measures to guarantee decent working conditions to workers”. In order to encourage
platforms to adhere to such charters, it was specified that the elements in those charters
could not be used as a basis for the legal re-classification of the relationship between
independent contractors as a work contract. This amendment, which had been adopted, was
then censured by the Constitutional Councilxxvi, which considered that it was a “legislative
rider”xxvii. The provisions relating to these charters are currently integrated in the draft
mobility law currently being examined. More recently, an order by the social chamber of the
court of cassation of 28 November relating to the contractual relationship between a bike
delivery person and a platformxxviii, paves the way for the re-classification of certain platform
workers as employees. The relationship of subordination is clearly at the centre of debates
with non-negligible repercussions on occupational risk prevention. The social chamber of the
court of cassation considers that due to the use of a global positioning system enabling
permanent control of the platform over the activity of the deliveryperson and a series of
incentive and penalty tools to establish that the link of subordination cannot be discarded.
As Jean-Emmanuel Ray states in the daily Liaisons sociales magazinexxix, the economic
consequences of a requalification would be so considerable for platforms that they will
attempt to eliminate any traces of subordination. This could have the perverse effect
(regarding safety) of discouraging all prevention initiatives (for example, the supply of
protective equipment) by platforms.
Juicer: a new invisible job in the heart of cities
The arrival in large French cities, in particular Paris, of free floating electric scooter rental has
been accompanied by the emergence of a new professional activity performed mainly by
micro-entrepreneurs, that of “juicers”. These operators’ work consists in collecting scooters
at the end of the day (from 6:00 p.m.) to charge them (generally in their homes), and then
redeploying them in the city at dawn (before 7:00 a.m.). Collection and drop-off are guided
by global positioning functionalities on applications. For each scooter recharge, the operator
earns between €5 and €20. Seemingly trivial, this activity is quite interesting for OSH
professionals, since it reveals the extent to which health and safety concerns were neglected
in upstream reflections and emerge only once the activity is practiced in the field, as
highlighted by certain press articlesxxx. Some characteristics of this activity are of particular
concern:

 operators compete against each other, therefore, there are situations of conflict
(sometimes physical violence) to recover the scooters tracked by GPS;
 remuneration varies based on the location of the scooter. If it has been left in a place
hard to access, or closed, remuneration is greater. This could encourage juicers to
break certain rules, for example, to enter the courtyard of private buildings to pick up
scooters;
 work is done at night, involves a lot of driving and handling (each scooter weighs
between 12 and 20 kg) with a vehicle that is not necessarily adapted to that activity;
 a juicer can take up to 30 scooters to their home to charge them simultaneously,
which obviously poses the question of compliance of the electrical installation and
consequences in terms of a possible overload, which can cause a fire in a private
room full of batteries.
The occupational health and safety challenges clearly were not taken into account (or
willingly outsourced to independent contractors?) by these platforms. They emerge
suddenly once the activity has been deployed, and generally come to the fore when
operators “pick up the pieces” and eventually speak out.
The opioid epidemic in the USA (also) an occupational health and safety problem
The opioid crisis in the USA appeared in many press articles in 2018. Particular emphasis was
placed on the health dimensions of the problem, with a record number of 72,000 overdose
deaths (including two-thirds due to opioids) in 2017xxxi and their impacts on the labour
market, in which the participation rate (people with jobs and seeking jobs) has declined,
attributed by several economists to the consumption of opiates xxxii.
The occupational dimension of the problem is now emerging, as illustrated by an article in
the New York Times of 26 September 2018xxxiii. Describing the life of a bricklayer addicted to
drugs, it showcases the difficulties posed by this crisis to American employers. In a survey
conducted with employers of over 50 employees, 70% of them state that their business has
already been affected by problems related to drug consumption: absenteeism, positive
screening tests, accidents, overdosing at the workplace, etc. Only 13% state that they trust
their ability to detect high-risk consumption. Official statistics show an increase in mortal
overdoses in the workplace of almost 25% per year since 2010, reaching 217 deaths in 2016.
After the entertainment and the catering and hospitality sector, the building and public
works sector is affected the most. Apparently, 1.3% of workers in this sector are dependent
(twice as many as in the rest of the economy) and a quarter of overdose deaths at work
concerns construction workers. Occupational factors are not absent in the addiction process,
with some workers being prescribed opioid painkillers after an episode of back pain or
accidents and becoming dependent afterwards. The article also speaks of the occasional
voluntary blindness of certain employers, aware that a non-negligible portion of workers at
their worksites have consumption problems. By way of illustration, it gives the example of a
construction company whose head, driven by the overdose death of one of his employees at
a worksite, decided to reinforce his policy towards drugs, in particular through systematic
screening. This approach led him to fire 12 of his 50 employees. Lastly, the article shows that
certain professional organisations obtain promising results, thanks to prevention
programmes using several approaches: detection of course, but especially individual follow-
up and insertion through work.
Not at all on the same level as the American situation, consumption of opiates has however
increased in France, where they have become the leading cause of overdose.
FOR MORE INFORMATION

 Evolution in modes of production and occupational risks: a review on horizon


scanning in 2017. Hygiène et sécurité du travail, 2018, 251, pp. 108-115. Accessible
at: www.hst.fr (in French)
i
Available at www.inrs.fr/prospective
ii
i.e. approximately €31,250 (as at 10 July 2019). This price is that of a low-level model. The corresponding model is
currently not available for sale. Only the more sophisticated models are available, at a price of about €60,000.
iii
N.E. Boudette – Elon Musk’s mass-market gamble. The New York Times, 30 June 2018 See :
www.nytimes.com/2018/06/30/business/tesla-factory-musk.html
iv
For example, placing engines and batteries in the vehicle:
- For the previous models produced by Tesla, that required 14 operations and the involvement of 17 workers;
- - for the production of the model 3, the process was reduced o 5 operations, without any human
intervention.
v
A. Feitz – Tesla: les leçons d’une robotisation à outrance. Les Echos, 2 July 2018. See
www.lesechos.fr/02/07/2018/lesechos.fr.0301899478796_tesla---les-lecons-d-une-robotisation-a-outrance.htm
vi
W. Evans, A.J. Perry – Tesla says its factory is safer. But it left injuries off the books. Reveal (Center for Investigative
Reporting). See: www.revealnews.org/Article/tesla-says-its-factory-is-safer-but-it-left-injusries-off-the-books/
vii
N.E. Boudette – Tesla Factory Safety under Scrutiny after Worker is injured. The New York Times, 20 April 2018. See:
www.newyorktimes.com/2018/04/20/business/tesla-plant-safety.html
viii
Can be consulted at: twitter.com/elonmusk/status/984882630947753984
ix
See: https//media.ford.com/content/fordmedia/fna/us/en/news/2017/11/09/ford-exoskeleton-technology-
pilot.html
x
S. Seibt – Ford fait entrer des exosquelettes dans ses usines. France 24, 8 August 2018. See:
www.france24.com/fr/20180808-ford-exosquelettes-usines-automobile-sante-ouvrier-cout-technologie-robotique
xi
See: www.youtube.com/watch?v=YS1gEx2Selc
xii
They were designed by the company RB3D, specialised in cobotics.
xiii
M. Ryan, C. Metz, R. Taylor – They can grab, they can twist. Watertown Daily Times, 1 August 2018. See:
www.watertowndailytimes.com/national/they-can-grab-they-can-twist---and-soon-theyll-have-the-touch-20180801
xiv
A. Magone, T. Mazali – Voyage dans l’industrie du futur italienne. Transformations des organisations et du travail. La
Fabrique de l’industrie (Presses des Mines), 2018
xv
See: www.cnil.fr/fr/comment-permettre-lhomme-de-garder-la-main-rapport-sur-les-enjeux-ethiques-des-
algorithmes-et-de
xvi
See a graphic description at: www.youtube.com/watch?v=L7iLdQ8WK5A
xvii
See the group’s press release on the launch of the platform at: www.publicisgroupe.com/fr/news-fr/communiques-
de-presse/publicis-groupe-devoile-marcel
xviii
A. Bodescot, V. Guillemard – Les unisnes 4.0, nouvelles cibles privilégiées des cybercriminels. Le Figaro, 12 December
2018
xix
Nicole Periroth, Clifford Krauss – A cyberattack in Saudi Arabia had a deadly goal. Experts fear another try. The New
York Times, 15 March 2018. See: https://www.nytimes.com/2018/03/15/technology/saudi-arabia-hacks-
cyberattacks.html
xx
D.E. Sanger – Hack of Saudi Petrochemical Plant Was Coordinated From Russian Institute. The New York Times, 23
October 2018. See: www.nytimes.com/2018/10/23/us/politics/russian-hackers-saudi-chemical-plant.html
xxi
T. Mackie – New era of state sponsored hacking can turn oil rigs into bomb that can kill. L’Express, 18 February
2018. See: https://www.express.co.uk/news/world/920437/computer-hacker-cyber-hack-saudi-arabia-cyber-
criminals-oil-rigs
xxii
D.E. Sanger – Pentagon puts cyberwarriors on the offensive, increasing the risk of conflict. The New York Times, 18
June 2018. See: https://www.nytimes.com/2018/06/17/us/politics/cyber-command-trump.html
xxiii
D.E. Sanger – U.S. declines to sign declaration discouraging use of cyberattacks. The New York Times, 12 November
2018. See: https://www.nytimes.com/2018/11/12/us/politics/us-cyberattacks-declaration.html
xxiv
E. Baumgaertner – As D.I.Y. Gene Editing Gains Popularity, ‘Someone Is Going to Get Hurt’. The New York Times, 14
May 2018. See: https://www.nytimes.com/2018/05/14/science/biohackers-gene-editing-virus.html
xxv
“Avenir professionnel”, amendment No. 2072. See: www.assemblee-nationale.fr/amendements/1019/AN/2072.asp
xxvi
Decision No. 2018-769 DC of 4 September
xxvii
Measure introduced by a law through an amendment without any connection to the initial draft/proposal, in
disregard of the Constitution
xxviii
Court of Cassation (social chamber), 28 November 2018, appeal No. 17-20.079. See:
www.courdecassation.fr/jurisprudence_2/chambre_sociale_576/1737_28_40778.html
xxix
In: Liaisons sociales (daily), No. 17708, 6 December 2018, p.2-3
xxx
“On se tue la santé pour €50 par nuit”, José chargeur de trottinettes électriques. Ouest France, 25 October 2018.
See: https://www.ouest-france.fr/ile-de-france/paris-75000/se-tue-la-sante-pour-50-eu-par-nuit-jose-chargeur-de-
trottinette-electrique-6031657
xxxi
Les Etats-Unis tentent de réagir face à la crise des opioïdes. Le Monde, 15 October 2018. See:
https://www.lemonde.fr/sciences/article/2018/10/15/les-etats-unis-tentent-de-reagir-face-a-la-crise-des-
opioides_5369765_1650684.html
xxxii
La crise des opioïdes, un défi pour le marché du travail américain. Les Echos, 31 August 2018. See:
https://www.lesechos.fr/monde/etats-unis/la-crise-des-opioides-un-defi-pour-le-marche-du-travail-americain-137760
xxxiii
Workers overdose on the job, and employers struggle to respond. The New York Times, 21 September 2018. See:
https://www.nytimes.com/2018/09/21/business/economy/opioid-overdose-workplace.html

You might also like