Commonwealth Bank of Australia has been put on a blast after an employee, Kathryn Sullivan, accused the bank of sacking her and replacing her with a chatbot.
Sullivan, 63, had worked for the bank for 25 years of her career before the incident.
She worked as a teller and helped to train a chatbot to handle her tasks. Then, in July, the management of the bank sent her packing and used the chatbot to do her job fully.
Sullivan said she was supportive of new technology introduced to enhance their processes in the bank.
She lamented that the decision of the bank shocked her and her colleagues. The sacked employee said she felt like the bank’s management saw them as numbers and not real human beings.
To complicate her matter, the bank, according to her, refused to communicate with her for eight days.
Some of the last things she did before being sent off included scripting and testing chatbot responses for the bank’s Bumblebee AI.
She managed the trialling of the technology with live customers. When the chatbot did not respond to a customer, she would step in to attend to the customer.
The sacked employee lamented that, unknown to her, she was training a chatbot that took her job.
Sullivan called for a regulation to ensure that AI does not take the place of human beings in the workplace.
CBA later reached out to Sullivan to resume her job after the chatbot failed to fully satisfy customers.
The bank said it made a mistake in assessing the roles. According to the bank, calls from dissatisfied customers spiked after the employees were laid off.
The management offered the affected staff their jobs back, but Sullivan decided not to return.
She said the role they gave her was not exactly what she did before the sack. The sacked employee expressed fears that the new role offered no real security.
The bank’s spokesperson said the initial assessment that 45 roles were not required failed to adequately consider all relevant business considerations.
He said the bank had apologised to the affected employees. The official also disclosed that efforts were ongoing to review the internal processes that led to the error.
