Forum

Conversational AI or Natural Language Processing

Kurt Van der Veken

29 August 2018
9 comments
Total votes: 1

Language processing is one of the main domains in AI, which is expected next to Machine Learning to have a huge impact in organisations and society as a whole in the years to come.
I’ll refer to this domain as conversational AI throughout this post.

Examples of conversational AI which we already encounter day to day are Siri, Google assistant, Alexa. These examples showcase the impressive capabilities.

Other examples of conversational AI are chatbots on webpages of commercial shops.
These chatbots show that conversational AI is still not that mature yet. Chatbots seems to be limited to a small predefined set of actions or responses and have a limited understanding of questions being asked and need to hand-over to a human agent relatively quickly.
This is my personal experience, but I heard similar experience from colleagues and friends.

If chatbots evolve to virtual service agents (chat & voice) that can answer more complex questions and initiate actions autonomously this will have a huge impact on the customer experience and also on the workforce and required skills of service departments.

So related to this topic I have a number of questions, which could potentially provide inspiration for research but on short term I would like to get the opinions of the forum members.

When will mature conversational AI technology become more mainstream and available for use within organisations?
- When can we expect full language support for native languages?
- What are other challenges that are preventing the technology to become more mainstream?
When will organisations be able to adapt to conversational AI technology?
- Are the skills of the workforce the main issue?
- Do they have the capabilities to integrate existing systems with a new digital channel like a virtual agent of chatbot?
- What are the other challenges that organisations face in the adoption of conversational AI?
When conversational AI technology matures and adoption grows what will be the impact on the workforce like service agents?
- What will be the impact on knowledge workers in general on the long-term?

I'm looking forward to your comments, ideas and answers!

Tags:

Conversational AI Chatbots NLP Natural Language Processing

Comments

Amir Sabirovic,

04/09/2018 14:26

When will mature conversational AI technology become more mainstream and available for use within organisations?
- When can we expect full language support for native languages?
  - Not any time soon. The complexity of natural language, syntax, negations and meaning are too huge to generalise into one solution. Instead, you already have vertical solutions which handle one task at the time. I assume English, Spanish, French, Chinese and Japanese will have mainstream solutions. All other languages will and are left behind because they don't have the substance to justify the business case.
- What are other challenges that are preventing the technology to become more mainstream?
  - The lack of process and domain knowledge within organisations which are necessary for the development of any kind of AI solutions which replaces or automates a certain task in a process.
When will organisations be able to adapt to conversational AI technology?
- When we have the ontology of each and every market, including slang listed including all synonyms and meaning (per language).
  - Are the skills of the workforce the main issue?
    - Yes and no, the workforce is not able to utilise the possibilities of technologies which enable the automation of routine tasks due to lack of knowledge. On the other hand, the same workforce is not able to describe the processes and data streams of their tasks.
  - Do they have the capabilities to integrate existing systems with a new digital channel like a virtual agent of chatbot?
    - No, I believe existing organisations (pre-2000) are aware of the possibilities but have to big legacy soup they can't get rid off. The existing systems are outdated and in some cases obsolete.
  - What are the other challenges that organisations face in the adoption of conversational AI?
    - Defining the approach?
      - Do we have a problem?
        What kind or problem do we have?
        Strategic
        Tactical
        Operational
        What technology is the right solution?
        Why?
When conversational AI technology matures and adoption grows what will be the impact on the workforce like service agents?
- What will be the impact on knowledge workers in general on the long-term?
  - If and when it matures we will have a personal assistant which will do all kinds of boring and nog cognitive challenging tasks for us. So if done correctly we will surpass a huge amount of unnecessary processes. Such as apointment making (you will grant access to other assistants and vice versa), booking a room etcetera. So all information services will be a push of a button. But is this likely to happen 50/50...

I'm open for further discussion or comment :)

Alexander LÖSER,

03/09/2018 19:58

Dear Kurt,

I like your examples. However, most of them may "support" legal entities, which contribute to the B2C platform economy outside of the EU, such as Google, Amazon, Alibaba, WeChat etc.

I'm curious to learn more about examples supporting EUish scenarios, such as supporting B2B industry related or even health oriented platforms. Is your team aware of such scearios for chatbots?

Best's

Alexander

Kurt Van der Veken,

04/09/2018 09:11

Hi Alexander,

In Europe there are local players which offer chatbot or conversational AI functionality usually they build on-top of cloud services provided by Google and Microsoft. Some local players develop their own NLP and speech2text components for local languages that can be used in B2B. Unfortunuatly I have no experience for health oriented chatbot platforms.

Kind regards,

Kurt

12/09/2018 12:35

Hi Kurt,

could you please name some of the players that might provide their own NLG and intent engines? For example, in Berlin, it would be RASA, which might move it's HQ relocate to Sillicon Valley in 2019.

13/09/2018 15:59

Some local players which offer NLP engines;

21/09/2018 19:41

Very nice. However, they all seem to focus on B2C, which will be likely dominated by US and China-based plattforms. Any idea on other domains, besides ada and babylon for health care?

Bjoern Juretzki,

07/09/2018 19:23

Dear Alexander,

One recent example of a health-oriented chatbot is Babylon's systems for the British NHS: https://techcrunch.com/2017/01/04/babylon-health-partners-with-uks-nhs-to-replace-telephone-helpline-with-ai-powered-chatbot/?guccounter=1

"Babylon Health, the U.K. startup that offers a digital healthcare app via a mixture of artificial intelligence (AI) and video and text consultations with doctors and specialists, has scored a potentially significant trial with the National Health Service (NHS).

Working with a number of health authorities in London, Babylon will begin a six month trial starting at the end of January to offer its AI-powered chatbot ‘triage’ service as an alternative to the NHS’s 111 telephone helpline that patients call to get healthcare advice and be directed to local and out-of-hours medical services."

Regards,

Bjoern

12/09/2018 12:32

Thank you very much Bjoern, we also work with the competitor of Babylon here in Berlin, called ADA.com. They focus on nearly the same idea as Babylon.

Does anyone know also nice application cases for conversational ai in industry (beyond call center support), in particluar B2B cases, or other sectors, besides of the obvious B2C cases already covered by Microsoft, Google, Facebook, Amazon, Alibaba, Tencent and alike?

Moreover, datasets seem to be a critical problem. Most research data sets focus more or less on Q&A.What really happens in industry seems to focus more on states. Malubaa published Frames, a data set for a B2C scenario on booking trips. However, Maluuba was aquired by Microsoft last year. As mentioned earlier, I see here for Europe more opportunities in Health and Industry. Are you aware of datasets for those domains?

Best's Alexander

02/10/2018 16:37

Dear Alex,

You mean datasets with spoken language beyond Q&A/call centers that could be used for the development of AI-powered chatbots?

European AI Alliance

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Forum

Conversational AI or Natural Language Processing

Comments

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Share this:

Comments