Conf Day 1. 01 Dec 2023 . " Generative AI as your Faciliative co pilot " by Mrunal Lamge & Subbu

 GENERATIVE AI by Mrunal Lamge and Subbu. 

This turned out to be an extremely exciting session. In my table were Nasreen Khan, the Russian , some other guy . Nasreen, Russian had laptops . Facilitators wanted at least one laptop in each table so that, they can type in ' prompts ' into their laptops, using mobile internet. All of them did. Even Madhujit Singh brought her laptop. Had I taken my laptop instead of fast scribbling into my notebook, I would have taken a lot of notes , a lot more notes, from the 6 sessions I attended. A small regret . Nevertheless. 

Madhujit Singh chose this session . So did Larry alias Lawrence Philbrook. It was great . They asked us to use Google Bard and Chat GPT as well. I taught Nasreen, the prompts and the regenerate . 

I need to work on these prompts. Deepak Verma was also there. He said he uses chat GPT to decide the title of his programs for his clients, namely, Microsoft and Oracle. Like ' give me 10 different titles to this program " and he uses a combination . 

The discussion revolved around - human element in using AI and the ethical aspect of it as well. 

Pitfalls of over dependance on AI were also discussed. 

-------------------------------------------------------------------------------

FOR DOCUMENTING . 

" GENERATIVE AI as your facilitative Copilot " by Mrunaal Lamge & Subbu .01 Dec 2023. Hyderabad .  Hotel. 

Suggestion : Before you launch this session in your hub, please ask all registered participants to come with their laptops to the workshop . They need to connect with their Wi Fi hotspot to input some prompts into Generative AI . 

They can do so from mobiles also but a laptop would be more effective. This , we learned from experience  on 01 Dec 2023 at Hyderabad IAF Asia Conference. 

Started with introduction .  Ways of working explained. Also the session plan clearly drawn out. Explained by facilitators. 

Ways of working was explained by Mrunaal and Subbu and they asked us if we want to add anything. 

Session plan was clearly explained. 

BEGINNING , DISCUSSION ROUND 1 : 

At that time, around 01 Dec 2023, Google Gemini as now it's called, was called ' Google Bard' and MS Copilot was called ' Bing Chat ' . 

So, our facilitators asked us about what we already know about Generative AI and Large Language Models like Chat GPT and asked us if we have used it and to what extent. 

Asked us how many of us used MS Copilot ( Bing Chat at that time ), Chat GPT & Google Gemini ( Bard at that time ) at least once and what was our experience like and for what purpose and what would be our advice to the new users of Large Language Models. 

We all chipped in with our responses. 

Different participants had different levels of exposure and actual usage. Some were really professionals in giving prompts and getting what they wanted, and some of us were just wetting our feet . Some others were only aware of it, heard about it and did not actually use it. 

Some of us used them for searching training content, some, for planning a facilitation session , some , for planning facilitation methods , activities etc and some for creating images, and some for creating a ppt . 

One participant told that, he uses chat GPT to give an attractive title to his training sessions . He asks Chat GPT to come up with 10 titles for his program and he uses a mix of them to create a synthesis and a new title. 

Testing the existing knowledge and experience of the participants helped the facilitators on how to go about it and from where to pick up .  It also helped us with a great peer to peer learning that was deep  , immersive and insightful. 

DISCUSSION  ROUND 2 : Q4D ( Question for Discussion ) 

In the second round of discussion , Subbu / Mrunaal asked us, " What was your last prompt into Generative AI ( input ) ? And what was the output, as far as you remember ? 

For 20 minutes this led to a very interesting discussion on the last prompts .  We searched for content , activities, group discussion questions , images, PPT s, and for concepts, delivery methodologies. 

It surprised me that, some facilitators who used Chat GPT quite regularly, did not know how to ' regenerate response ' and where the ' Regenerate Icon' was ! We helped one another. 

SECTION 3 :  PROMPT EXERCISES. 

This was the MEAT of the program. The most important , which led to activity based , experiential learning and discovery learning ! 

Here is where we made a mistake. In our table, 6 to 7 people sat. Only Madhujit Singh, Russian Andrei Kolesnikov & Nasreen Khan had laptops. We didn't have. We could type the Prompts into our mobiles easily. So we watched them doing that. 

In this kind of activity based training there are 2 suggestions, based on my experience. 

(1) Every participant should have laptop and connect to internet with Wi Fi hot spot. or 

(2) Create a WhatsApp group for all participants one day in advance so that, you can post your prompt into the group during training.  They can copy paste the same into different Large Lanague Models / Generative AI chatbots. 

PROMPT 1  : The prompt Generation Chalenge 1 . 

Go through the following client ask & work with a Generative AI tool of your choice and come up with a draft program design. 

Day 1 : Initial Ask. Client : Tech innovations Inc . 

Ask : Tech Innovations approaches you with a seemingly straightforward  request . 

" We need a team-building workshop for our Development Teams to enhance collaboration and creativity . " 

Your understanding : Day 1 : 

You learn that, the client is looking for a workshop on team building and creativity among their development teams. Initial discussions reveal a desire for engaging activities and strategies to boost collaboration among the Technical teams. Two teams of 25 members each, workshops to be done in person. 

( End of Prompt 1 ) 

There were 6 or 7 tables, each table had 2 or 3 laptops . Esp outstation delegates. So, they entered these prompts into Chat GPT , Google Gemini & MS Co pilot. 

Got the responses. ( 5 minutes )

Divergence activity is over. Emergence is over. 

Table Discussions : We entered the given prompts to LLM s and discussed different responses we got, among our table mates. We also clicked on ' Regenerate Response ' to Chat GPT . Got the best of the responses and collated and consolidated into one response. 

Convergence activity over. 

Team presentations from each table. And questions from other tables. Some discussion . 

5 minutes for each table and there were 6 or 7 tables .  

For a lot of facilitators who were not familiar with LLM s, this was activity based experiential learning and discovery learning. 

Once the first round is over, we were asked to write our key learning in one sentence and post on the sticky wall. 

Immediately, without much gap ( perhaps due to time constraint ) , round 2 started. 

PROMPT CHALLENGE 2 : 

Unveiling Complexity 

New Insight ( Day 2 ) As you dive deeper into the requirements of the client, you discover that, your client is undergoing a major organizational restructuring ! The development teams are not only dealing with interpersonal challenges but are also grappling with a shift in project priorities and team compositions. 

Your updated understanding ( Day 2 ) 

The ask now evolved to , not only focus on team building and creativity but also to include elements of change management and adapting to restructured organization . 

You gain insights into specific challenges related to restructuring, such as potential resistance to change and the need for clear communication . Further, you also understand that, one team is tenured, while the other is fairly new to the organization . 

Task : Update prompt to the Generative AI tool and modify the program design. 

This time too. we didn't type the given prompt as it is, into the LLM. We edited, used our own language, made some changes and entered the prompt. 

As the LLM threw up program design, we had table discussions and made the required changes. All of us gave inputs. This was similar to World Cafe method. Not like Hell Frame game. For, there were no conversations between tables and table members did not move from table to table. 

So, there was human element in modifying the prompt ( all tables, all members with laptops made changes in the given prompt as per their own understanding ) , also collating the LLM responses we got and finalized the program design after table discussions. 

The program was very fast paced due to time constraints and the table discussion time, for the second prompt, we felt was not enough. 

Second round : Team presentations, questions by other tables. It was very very interesting and resulted in immersive peer learning. 

Question - answer session between tables. 

THRID PROMPT CHALLENGE : 

Hidden Layers Uncovered. 

SITUATION : ( Situation, our understanding was given to us and we need to convert this into a good prompt, after table discussion and enter the prompt into LLM) 

In further discussion, you learn from the client that, restructuring is not just a response to market conditions and trends but is driven by a large tech company of a start up ! 

The acquisition involves, integration of diverse teams with varying work cultures, creating the need for the workshop to address not only internal collaboration challenges but blending of cultures ! 

YOUR FINAL UNDERSTANDING ( DAY 3 ) : 

Their ask & you task, now fully coloured with complexity , reveals that the team building workshop is intricately linked to the larger context of an acquisition ! 

The facilitator understands the need to address cultural integration, manage potential conflicts arising from the merger, foster a unified , innovative new culture within the new , combined entity ! 

The ' Tenured group ' is from the old ' acquiring ' organization and the new group is from the new , acquired start up entity that is being merged with the larger firm. 

Time duration : The teams have only one day , to spare, for this workshop ! 

Your task : Update this prompt to the Generative AI tool and modify the facilitation program design. 

Table Discussion ; We took 5 to 10 minutes to create a good prompt. Entered into Chat GPT . 

Then, copy pasted the response to word doc. Did 2 more rounds of ' regenerate ' . Different responses surfaced. 

We merged points and made a brief team presentation document that can be presented in 3 minutes. Due to time constraints. 

Followed by Team Presentations, Q&A between tables. 

Since all participants had varying degrees of exposure, experience and expertise with Generative AI and LLM s, the conversations were very animated and vibrant. 

Session Wrap up : With reflection of learnings, most important points learned, we wrote on sticky notes and pasted on the sticky wall. ( Divergence ) 

' Themes Emerged ' and facilitators , with our consultation, re-arranged them according to emerged themes. 

Kolb's Learning Cycle, from ' Concrete Experience, -> Reflection --> Active experimentation -->  Abstract Conceptualization - have been implemented to perfection . 

It has been a great learning experience to all.