Comparing Bing Chat , Open AI chat GPt and Google bard: Chatting with BOB

First half of 2023 has been exciting month in terms of further developments in the AI ecosystem, while Meta launched new Llama version. MS integrated Open AIs chat GPT capabilities into its Azure stack of services as well as power apps, flows and AI generator. Microsoft also launched Bing Chat public and Enterprise version rapidly scaling GPT 4 based chat searches to an enterprise grade tool with many additional capabilities. Google Bard which works on PaLM - 2  is also improving gradually , I decided to compare Bing , Open AI chat GPT and Bard (BOB) form an end user perspective with publicly available versions (Not Enterprise versions). For the purpose of this comparison I have ignored any open AI chat GPT plug ins which may extend its functionalities and allow it to perform some more tasks.   GUI:- While Open AI GPT has still been running on same GUI from last 6 months or so its other counterparts have not only done some improvements but also used their ecosystems to integrate their search Bots with it Open AI CHAT GPT has prompt input, output area as well as historical view. Whereas if we look at Bing Chat it has added image and voice inputs with text for prompt input . It provides latest web indexed up to date information post searching from internet and citations as well as follow up prompt suggestions related to topic. One of the key drawback of Bing chat is that its currently locked in with MS edge and doesn’t work with any other browser. Bard is also well integrated with google search. Its prompt input supports both text and voice (No image yet) but you can provide an image of video for it to analyse via URL for example you can directly give it a you tube video link and ask Bard to explain same .Once the output is provided there is option to convert response text to voice as well as share or google and provide feedback. The output of each prompt is provided in 3 draft variants and any one of those can be further explored.  Up to date information: Both Bing chat and Google Bard use indexed internet UpToDate information to provide prompt response whereas open AI has still been restricted to 2021 data making Bing and Bard immediate first choice for users to get their answers based on live information.   Citations: One of the key struggle for organizations and users developing AI use cases is to get the clarity and source of information behind the output provided and Bing and Bard has now taken next step by providing link to citations of their source information whereas in Open AI chat users are still left wondering form which exact source the information has been provided.   Custom doc training: Bing and Bard have provided easy way to provide response on custom user docs , Bing can access links to a web page or a document which Is opened in MS edge  and provide summarization , answers on those document which is also possible for Bard but same is not an off the shelf functionality with Open AI chat GPT.   Image and Speech handling: While Open AI GPT is still a text-based interface even after having a sibling Dall E available for image generation its synergies have not been used. Microsoft Bing has combined both images and text and not only it can recognize images but also use Dall-E to generate images for various prompts. Bing is also able to understand speech prompts. Google bard has also extended its functionalities to speech where it is able to understand speech prompt and also read out the outcome to users. These functionalities represent true inclusiveness for AI which will enable Bing and Bard to reach larger section of users.   speed & Temperature control: All 3 chats support temperature control helping response to be either more factual or more creative. Since Bing and Bard are actually using internet search as well they seem a bit slower then Open AI GPT which starts generating response as soon as prompt is submitted.   Summary: Overall Microsoft has leaped ahead in using its mighty ecosystem and integrating GPT4 based bing chat with its various platforms thus creating a perfect package for organizations which can be adopted rapidly without much work for their custom data training. Its key to highlight while GPT 4 has been found to be better then GPT 3.5 , recent study by UC Berkley and Stanford which compared two lates versions of GPT 4 released between March and June have raised concern about GPT 4s performance and accuracy for mathematical calculations . Google Bard has also been catching up rapidly and available for enterprise use by Google workspace. Open AI GPT has been released for commercial use with various plans and is open for enterprise use either directly or via API integration.

 Keeping an Eye on AI

Recent announcements and developments in the field of Generative AI has triggered a race to “AI first” systems. Within 2 months over 100 million users rushed to experiment with chat GPT and all large product organizations have been making announcements about new products, use cases and LLMs. This creates a ripple effect for business and operations for faster adoption and especially fear of missing out. What also need to evolve at same speed and urgency is local and global governance for these AI systems and use cases at both technological as well as ethical level. While entire concept of AI is being independent in creating, learning, and taking decision, we have sufficient and significant examples to indicate that we need to keep an eye on AI systems and design them in such a way that enables monitoring of various key parameters like efficacy, bias, usage of data and adherence to various legal frameworks.

Recent announcements and developments in the field of Generative AI has triggered a race to “AI first” systems. Within 2 months over 100 million users rushed to experiment with chat GPT and all large product organizations have been making announcements about new products, use cases and LLMs. This creates a ripple effect for business and operations for faster adoption and especially fear of missing out. What also need to evolve at same speed and urgency is local and global governance for these AI systems and use cases at both technological as well as ethical level. While entire concept of AI is being independent in creating, learning, and taking decision, we have sufficient and significant examples to indicate that we need to keep an eye on AI systems and design them in such a way that enables monitoring of various key parameters like efficacy, bias, usage of data and adherence to various legal frameworks.   •             Hallucinations handling :  Its not unlikely for humans to speculate or make statements based on assumptions. With human like creativity its obvious that AI has also learnt to provide made up answers when it doesn’t have needed facts and data on the topic, this is called hallucination. In near future there will be more development to handle hallucination, but two things are key here first of all to train AI model with as much data as possible and second to provide users a feedback options so that they can report an hallucination of the system to owners for corrective actions and interventions.   •             Watermarking:  With AI creating human like artifacts its becoming increasingly important to differentiate what’s generated by AI and what’s not hence watermarking the content created by AI should be one of the basic and standard principles.     •             Sustainability:  In an interview in Q1 2023 Sam Altman, Open AI CEO had mentioned that cost of cloud infrastructure training and running chat GPT is eye watering. Open AI and all such organizations are using thousands of chips and other hardware (directly or indirectly) as well as energy which have significant and probably equally eye watering impact to environment ranging from carbon footprint to e waste generated with out of use hardware. This makes its critical that all AI use cases in business and IT consider possible impact on sustainability vs the benefit such AI systems will generate. A technological eco system need to be evolved which can help us in better review and decision making of environmental impact of any It system.   •             Performance Monitoring:  At the current fast paced AI adoption into business at times the ultimate business functionality is of paramount importance but with that we also need a comprehensive approach for performance monitoring and benchmarking tailored to each system. While its good to rely on product vendors promises, we need to embed our own performance review and benchmarking which will lead to overall efficacy improvement for system and models.   •             Human in loop: At the early stage of AI adoption for any system and process we must keep human in loop to accept or reject the outcome generated by AI. This will help in setting up accountability on the team or organization for the decisions taken by AI .   •             Transparency: We discussed watermarking in this document but that’s invisible to human eyes and mostly detected by system whereas its critical that all users know and understand what part of their decision or outcome they have been handed over is coming from AI hence systems need to be designed for giving clear warnings and disclaimer to business users and also how has that been arrived

.

Can a Lady Speak ?

 

Once upon a time in there was an analytical platform called chankaya  to provide critical insight to business users , the massive platform had more than 25 business critical services hosted with thousands of dashboard providing real time as well as historical view . There were three leaders and respective teams supporting the ecosystem Brahma Kumar was responsible for all innovation from hardware to semantic layers , he ensured the internal machineries of system ran smoothly abstracting all the complexities from end users , he kept performing system upgrade bringing new and improved infrastructure as well as resolving old issues with system . He had two more counterparts , Vishnu Vardhan responsible for BAU activities of Chanakya who had teams of Application support consultants taking care of daily data load , front end and user issues . The teams of Brahma and Vishnu were highly technical and SMEs in their areas hence there was another small team present in ecosystem which was lead by Prakriti , Prakriti and her team under stood business context of system and report and they provided an interface between technical teams and Business users .

Business users looked up to prakriti for all communication and decision related Chanakya’s availability.  In 2016 Brahma suggested a major version upgrade to Chanakya’s database which was nearing end of life in few months . The business case was created and approved by needed stakeholders , Vishnu lined up his teams for all necessary support , testing , reschedules whereas Prakriti worked with business leaders to understand bets suited weekend where a downtime of 12 hours can be taken and system can be upgraded , the idea was to upgrade system on Saturday , ran needed loads on Sunday and let system be available on Monday for users to use . The D day arrived , by then all teams had rehearsed the plan and executed same in non production environments , there was good confidence that activity will be closed within time .

Chanakya was taken down , old data base backed up and restored into a newly built database environment , user permissions migrated and DNS names switched some of the initial tests were encouraging where technical teams could connect to data base and run query , see data . The ultimate litmus test was to trigger data loads , these data load jobs were veins of chankaya transporting the life blood – data to all the front end dashboards . Initial data loads refused to trigger from new data base , team assumed basic problems of access controls , job paths , source connection , firewall ports and checked one by one but none had shown any issue . Brahma and his team was already on the job from Saturday morning till Sunday morning , 12 hours past deadline of releasing the system , Vishnu was keeping close eye and his team was running all checks and supports both knew that a lot of effort has been spent on the project so far and it must get forward fixed to avoid restarting the exercise which will take months .

Prakriti on the other hand were worried that we have now crossed Sunday morning , a very thin window of running the data loads and making system available to business is left , if we continue fixing the issue , there will be chaos on Monday morning across multiple business functions . A checkpoint call was set up at 10:00 AM on Sunday to discuss next action items , when Prakriti joined Brahma was already updating on list of actions his teem took so far and he thought may be a full reboot of new data base and clearing the log once again could do the trick , the heated discussion on technicalities , steps involved went forward and backward between Brahma and Vishnu . After listening this for a while prakriti spoke – “can a lady speak”?

Call went silent which was broken by prakriti’s voice , “ A functional system even if old infrastructure is worth millions whereas a non functional system on upgraded infrastructure wont be of any value to us or business “  while I understand all of our effort in coming to this point but our commitment to business was to return a functional system on Monday which I am determined to do , we will need to revisit our approach and perform this upgrade in near future , I will make a case with business and get us another slot in next quarter by when our expert teams would have identified and fixed the issue .

Both Brahma and Vishnu could see the logic and agreed with heavy heart .

“can a leady speak ? “ still remains a key question in a cosmos where nature has inherently created balance with 50% women , this question had large social relevance , it took long time in many societies to let lady speak on key decisions and yet “can a lady speak ?” still remains a question in many of the developing societies .

Can a lady speak in matter of house hold finances, Can a lady speak in deciding future of her children in fact can a lady speak in deciding her own future ?

This inherent barrier has automatically crept into corporate world , in 21^st century it will be just to say it happens unintentionally and subconsciously unlike in previous centuries .

Ultimately lady will speak , woman will need to find her voice and speak assertively , some times it may need asking a dramatic question “can a lady speak?” but it should be a collective effort on all of us that no one needs to ask that question , we ourselves need to go around asking opinion from all those who missed out speaking , some times voices need enablement and support !.

The Go Live

Week had just started , On Tuesday itself It was proving to be hectic week for Venkat. There were continuous meetings and report preparation for newly live HR application . As per normal change and release process this was sent to production over medium release weekend after completing all formalities and Monday was the first day when employees and HR of organization started using it , so far they were used to a legacy system which had less functionalities and limited usage hence many of the activities were performed into various different applications or even out side system by mail or hardcopies , New configurable system definitely had an edge over older ones as it was true representation of Global HR best practices and had lot more automated functions then actually needed , some even futuristic which organizations can activate whenever they need .

Venkat is operations manager and handles very large portfolio of IT support and services , after landing into RUN mode this application was his responsibility , though there was still post go live support active by Innovation team lead by Prajapati who was actively involved in managing the crisis , there was already mails flowing from HR leadership to provide quick and efficient support , VP of HR even provided a deadline to provide solution within a week or HR would move back old legacy system which was still fall back option as Business continuity and had to be decommissioned once new system is completely adopted across organization .

Since Venkat and team had good understanding of user behavior , issues and expectation from previous system , Management had directed RUN team to support project in this crisis . Venkat himself was very keen to engage in early stages for this strategic HR solution , He had also been active participants in all gates meeting and tried to put all kind of checks to ensure delivery of a robust and resilient service however he was now regretting not pushing the case for performance testing hard enough , in one of the gate meeting when Prajapati proposed skipping performance testing as there was not sufficient budget to establish a production like environment and go for automated performance tetsing simulating user concurrencies , multiple parallel processing etc , Venkat did raise an objection mentioning Performance testing is crucial to understand basics thresholds which was supported and agreed by all key stakeholders in meeting . Taking a note of this condition Prajapati mentioned He would discuss with project implementation partners and product vendor to see how can we assure performance of the application , In subsequent meetings Prajapati produced an assessment completed by product Vendor which assured basic performance levels based on system capacity , number of users etc .

Venkat accepted this report and it analysis however post implementation System performance seems to be the most critical issue , New HR system had many advanced features and users were complaining these features take minutes to load and complete . On the other side while older system had limitation of number of employees associated with a manager , new one had no such limitation which enabled users to assign many employees with Manager but it made activities extremely slow , though product vendor had given a much higher threshold for assignment but system was showing issues at half of the threshold itself . Another challenges seems to be older version of  internet explores and laptop configurations , Change management had not been able to asses that such advanced features may need some uplift of users laptops .  All these issues have created many incidents in the first day itself and huge frustration in Users

The Fat Indian Bonus

Unlike rest of the world where bonus is given at the end of financial year , In india bonus is tightly linked to the biggest festival of the year - Deewali , on which most of the Indian Poor/middle/upper class outspend from thier budget to buy any and everything .

In 2014 , a Diamond exporter of Gujrat state savji bhai Dholakia , surprised its employees as well as entire nation by giving away 2 BHK flats , Cars and Jewelries worth 50 cr to one fifth of its employees ( near 1200) , as of now diamond cutting is not a very organized sector in India but Savji bhai has been giving all the facilities to his employees which are found in organized sectors , in fact he is going one step ahead and adding to this certain items which are core to India values and families like sending parents of employees on pilgrimage .

What stands out that Savji bhai has built great deal of understanding on the needs of employees and giving bonus suitable to such needs , he also understands , for Indian employees their family / parents . He has also kept on increasing the bonus share for employees with companies revenue and profit . by doing this Savji and his organization has definitely become best employer in diamond export industry

Crisis Day

It was business as usual one day and then it happened , didn't start exactly as Jungle fire but rather slowly , Key application used by many countries in a region had decided to take it easy , gradually users started complaining that application is not responding as fast as it used be , for some it wasn't even responding . Business panicked on possible loss of time and not able to move on with routine activities , multiple high incidents and a final blow came as an urgent incident - The moment of truth had arrived , formal P1 declared that means almost every one need to jump in to the crisis call , application support , Application/DB administrators , Managers , product SMEs.


Initially everything was a suspect , All teams started coming out with their analysis but seems no problem identified for funny system behavior, temperature was rising in conference , Business was waiting to move on in such case responsibility landed with product vendor , they have designed it and they would know it for sure and as always they were full of suggestions , we can try this or that , issue seems to be here or little there .


IT realized it was better to try something and move forward rather than not trying any thing so they started with options , subsequent restarts but it did not help so it was time to move on with second option , meanwhile looking at criticality of the issue some more resources pumped in hoping to revive system , that seems to be working for a while.


Teams went back to check further , monitoring on all parameter was put in place meanwhile application kept showing its mood swings and one day it again became business as usual , problem stopped as abruptly as it started  , some had clue that executed actions may have improved the condition only hoping that it would not happen again

Let them Know

A mega event was nearing and organisation was preparing with full thrust , recently developed marketing app to also be used in this event for first time , business was nervous about any issues , its usability , effective support . Constant pressure was built on manager of IT service provider who was supporting this app , every time he visisted his customers in IT/Business , people reminded him how crucial the upcoming event is , mails were started flowing highlighting the situation . IT team and managers were very confident knowing it was a very small app and they have some of the really best SMEs working with this customer for years , they are available and will handle any situation promptly , same was also informed to the the customer but panick continued ..


Support manager understood that emotions can be calmed by showing a structured plan , he sat down with team and developed a shift wise plan for the duration of the event , created a daily report template to be circulated , established escalation structure and wrote it down in the form of document , most of it was already in place as part of application support but team carved out a small support plan very specific to this application , same was submitted to to customer IT and business .
They could now see what all steps will be taken , who can be contacted in hour of crisis and above all thier vendor team is ready to stretch equally to make it a success , they are a stakeholder with written commitment in the form of a plan , emotions calmed down , all aspects were covered.

During event , support team adhered to the plan , there were no issue reported and yet they kept sending thier daily report template . For minor requests also resolution was provided in urgent manner . After successful finish of event Business team was deeply satisfied with IT support , large commendations followed with information on how beneficial new app was .

At times even though things are very obvious , we know that every thing is under control but same need to be communicated . In other times the simple communication may not work , we need to devise ways to comfort people , give them confidence and calm thier emotions.