Jorge Ramo López
Digital Workplace Specialist & TŌGŌ Product Owner at NTT DATA
Ignacio Barahona Esteban
AI Solutions Manager & Dolffia Product Owner at NTT DATA
How the new Gen AI models such as Microsoft Copilot are impacting the way we (will) work
As artificial intelligence (AI) continues to advance, so do the possibilities for its applications in various industries and sectors. One area in which AI is showing promising results is in the realm of generative models, which are capable of understand existing data and generating new based on patterns and rules.
Large technology leading companies have been working and investing on these technologies and now the first wave of tools are just reaching the market: ChatGPT, but also Bing Chat, Google Bard or Microsoft Copilot are some samples of tools based on Gen AI’s.
For sure, Generative AI’s has the potential to revolutionize the way we work and interact with public information but also inside a company across the employees. Recently, a case study of a major financial services firm has revealed the implementation of a GPT-3 based technology resulted in a 60% reduction in searching times for employees regarding complex financial reports. However, it also raises important ethical concerns that we must consider before a large adoption.
In this article, we will explore the ways in which generative AI models can improve the future of work such as enhanced productivity, improved communication, and the ability to access specific answers from complex repositories. On the way around, we will also delve into the challenges associated, including the potential for bias in training data. A quick use case will serve us to showcase how to get benefits from Gen AIs minimizing the associated risks.
Benefits of Gen AI’s in the digital workplace
First of all, let’s highlight some benefits we can obtain when applying this new generative AIs in the employee experience:
Enhanced productivity
Sometimes, as employees, we don’t even know what we are searching for. AI based tools help us to work more efficiently being able to find information as a real time collaborative chat, rather than doing a traditional keyword search. We can get more accurate answers quickly in the same way we are already testing with the new beta intranet search engines.
We can accurate answers in seconds regarding complex documents or event combining information from several sources or 3rd party systems.
Increased efficiency
The power of AI for the workplace has just been impacting in our daily work, but soon we will see a huge impact in the way employees do their everyday tasks. Having a tool that assists them in the least important aspects of the job (format, slide design…) will help them focus on the aspects that really add overall value.
Getting accurate answers from complex repositories
R&D, legal or technical documents may sometimes be hard to understand and require spending huge amounts of valuable time from employees. If we let the AI process the documents, we could ask more important questions in a matter of seconds: What reaction does this compound cause in this type of patient? What are the legal restrictions existing in this country regarding a certain field?
Employees can even ask in any language so the new Gen AI’s models understands the content and is able to translate the answer.
Improved communications
With these new AI powered capabilities, Content creators can define the headlines of a topic and let the AI fill the contents to fulfil your articles. Users can include internal or external information directly retrieved from the trending topics of the network, allowing them to focus on content curation such as selecting the right tone of voice for each collective.
Challenges ahead
However, the use of these technologies also requires to overcome some challenges. Let’s recap the main ones:
Privacy and confidentiality
Maybe the most important challenge associated with these technologies is the potential for sensitive information to be inadvertently shared or leaked through the platform. The AI’s use all available information to generate Knowledge, making difficult to discriminate which contents could be shared with a specific user. In an internet search engine, all the information crawled is public but in a digital workplace we need to deal with confidentiality levels and private contents for a specific groups of users. That’s why when introducing these new AIs in our digital workplace is necessary to define a secure and safe scope of information based on user groups and alert to our employ such a private community where only allowed users can read and share information.
Copyright issues
Since this technology is based on AI language models, it doesn’t create original content or generate new ideas. Instead, it uses existing language data and machine learning algorithms to generate responses to user inputs. The response may incorporate text or other content that is protected by copyright without permission from the owner, so we need to be very careful about how our employees will use or share the results provided.
Bias in the training data
AI generative models are trained on large datasets of text, which can introduce bias into the model. If the training data is biased in some way, such as containing stereotypes or discriminatory language, the model may generate biased outputs. This can have real-world consequences, such as reinforcing stereotypes or discrimination against certain groups. For example, a language model trained on data that disproportionately represents one group over another may generate text that is more favourable to the overrepresented group. To address this issue, it’s important to ensure that the training data is diverse and representative of the population.
Smart collaboration
Considering the benefits and the challenges of the new Gen Ai’s let’s think about a simple use case that can provide a lot of value: Smart Collaboration. Corporate Communities are the places where users share interests or information around a specific topic. Employees share documents or posts with their thoughts in a social way that represents a highly valuable source of Information and Knowledge. We can arrange communities thanks to TŌGŌ, our social digital workplace, and applying several Dolffia’s large language models (GPT-3, GPT3-5, GTP-4, bioGPT,..) and their pre-processor engine, users accessing the community can ask any question and get accurate answers.
Communities of interest could arrange around an specific Topic, Knowledge Area, Corporate information, R&D research or legal departments managing large contract information. Users can now easily answer the most complex questions regarding a set of documents or Knowledge base. We can even apply private communities so only a certain number of users can access some specific communities and knowledge.
And everything inside your company’s security compliance. No data will be extracted as both, TŌGŌ and Dolffia, can be deployed on the corporate tenant.
