REACH 477 - What are your views on the application of AI within Government services? What are some other ways of AI application that could help boost productivity of Government services and better benefit the public?
(SK)
19 Jul 2023 (10am - 7pm)
REACH
19/07/2023, 9:44 am - +REACH: *Dear contributors,*
Welcome back! π
⏰ We will be opening the chat from *10am to 7pm* today. ⏰
*House Rules (short version of our Terms of Use) to keep in mind:*
1. Be kind and respectful. We all want to be in a safe space to share our views.
2. Any and all threatening, abusive, vulgar or racially, religiously and ethnically objectionable content is prohibited.
3. Consider the quiet ones among us and give them a chance to comment.
4. No need to repeat your comment or in differnet forms (including caps) - we heard you loud and clear the first time.
5. Let's protect each other's privacy and keep contact details in this group what it should always be - confidential.
*Full set of Terms of Use: https://go.gov.sg/reach-whatsapp-terms*
We will strive to uphold these rules to ensure this is a safe space for all.
Please be assured that the points made by participants during the chat are aggregated and shared with relevant agencies.
The topic will be posted shortly.
Thank you.
Megan π
19/07/2023, 9:58 am - +REACH: π’ *Topic* π’
Some 4,000 civil servants are using a Government chatbot based on ChatGPT called Pair to assist them with writing, research, and coding, in a trial of artificial intelligence (AI) bots to boost productivity within the Government and better benefit the public.
A security arrangement between the Government and Azure OpenAI – an arm of Microsoft that offers coding and language AI models – also ensures that confidential information is kept private within ministries and unavailable to Microsoft.
π¬ *What are your views on the application of AI within Government services? What are some other ways of AI application that could help boost productivity of Government services and better benefit the public?*
Pair and an upgraded SupportGoWhere are part of the first tranche of AI use cases in the government. Pair will allow the secure and efficient use of large language models as a writing assistant within the government space, and also provides “a sparring partner for ideas”, he said.
The agency is also in the nascent stages of testing large language models to upgrade SupportGoWhere, a portal for locals to search and apply for government aid. Incorporating AI into SupportGoWhere will allow users to describe their problem in their own words and get a response on what assistance they can apply for and where to get it.
“If we get this approach right,” said Dr Janil, “citizens will be able to express their personal story without necessarily knowing how the schemes and support of the government are structured, which is something you still have to know today.”
Members of the public can take advantage of certain ePrimer modules on the Singapore Government Developer Portal to pick up these general concepts. (Link below - the last link)
ππ» *4,000 civil servants using Government Pair chatbot for writing, coding*
https://str.sg/iwvS
ππ» *How Singapore intends to harness AI for the public good*
https://str.sg/i3Ms
ππ» *Public officers can use ChatGPT and similar AI, but must take responsibility for their work: MCI*
https://str.sg/iJ8c
ππ» *Link to the Singapore Government Developer Portal*
https://www.developer.tech.gov.sg/
19/07/2023, 10:00 am - ~ REACH Phone 2 changed the group description
19/07/2023, 10:01 am - ~ REACH Phone 2 changed this group's settings to allow all participants to send messages to this group
----
19/07/2023, 10:02 am - +Brandon Lin: I’m glad to see that AI is being used and develop to harness the power of AI for the good of citizens to get necessarily help and support from the government especially a lot of the people are unaware of the helpful schemes they are entitled to
19/07/2023, 10:03 am - +Brandon Lin: It will be good if this development can be done in multiply languages so that people of different races can use their native language to input their scenarios
19/07/2023, 10:04 am - +Rama: It still has its risk but with adequate safeguards and due diligence can be regulated
19/07/2023, 10:04 am - +Brandon Lin: It will be also good to publish and promote awareness of such help to the public so more people are aware of it and uses it
19/07/2023, 10:19 am - +Frankie Wee: https://www.channelnewsasia.com/singapore/autonomous-buses-trial-jurong-island-science-park-430831
19/07/2023, 10:19 am - +Frankie Wee: SG future will have a School bus will be auto AI driver near their home kids pick point to drop point
19/07/2023, 10:22 am - +Frankie Wee: <Media omitted>
19/07/2023, 10:23 am - +Frankie Wee: But it can’t exceeded speed 60km/h
19/07/2023, 10:32 am - +Anselm: Good the government is investing into AI, autonomous vehicles, etc.
The challenge of doing more with less will increase into the future, for the government, especially when we see it from the prospect of a dwindling population
19/07/2023, 10:35 am - +Ken: It was trial very long finally ready after 5 to 6 years
19/07/2023, 10:35 am - +Ken: Oops 7 years ago already trial lol
19/07/2023, 11:29 am - +Rose M K Ho: This message was deleted
19/07/2023, 11:30 am - +Rose M K Ho: Questions are :
1. Who is regulating this (AI & its uses ) against abuse?
2. How will the public be protected against any potential abuse ie. misuse/loss of personal data, _mis-information_ , wrong advice/ medication ... and so on.
3. Will ther be a panel of staff to oversee this & monitor the system constantly to ensure smooth operations?
4. Is there any backup plan in case this system fails us , like a sudden glitch or 'power blackout' ?
5. If ther is, then has it (the backup) bn tested ?
I see this backup as _critical_ since government services r using it. Case in point - Immigration system breakdown at Woodlands not so long ago.
6. Wat if AI were applied to our defence system ? In a worst case scenario, no defence capability can be enabled due to power failure or the AI screw up everything! π€¦πΌ♀️
Dun get me wrong. I m all for tech advancements. But becoz I believe in it, all the more I fear for it - humans r getting too reliant on tech to get things done, not thinking of the consequences if these things fail one day.
Ther is a high tendency to take things for granted.
Hence, while I applaud the use of such high tech in our society, I m also sceptical of its applications -- _*ther must be proper policing procedures in place*_, overseen by someone in authority and proper monitoring , _*with backup in case of emergencies*_.
π€
19/07/2023, 11:31 am - +Frankie Wee: AI power backup 100% will work normal and abuse or violence law will take action against humanity.
19/07/2023, 11:32 am - +Frankie Wee: Human command to AI system biggest database into memory than human brain π§
19/07/2023, 11:35 am - +Rose M K Ho: Not good enuff, in my eyes.
Wat I mean is -- wat if AI messes up everything ... ie. Things dun work the way they r supposed to & people die as a result of these errors.
Can we go back to , say, _manual_ way of doing things, to put things right? -- tt is the kind of backup I m talking abt, not just power failure (again, human take for granted).
If got no power, then backup is to use candles, right?
So, if no AI, then how ? Ppl forget how to use their brains, is it ? π€ͺ
19/07/2023, 11:35 am - +Frankie Wee: Cybersecurity more power safeguard when AI will follow up law
19/07/2023, 11:40 am - +~L: Great to see the innovation and opportunity. There are different types of AI - Machine Leaning and Large Language Model are 2 examples commonly in use. I assume this initiative is Machine Learning. The computer can run many simulations and a human can then make the decision. This saves considerable time and resources.
19/07/2023, 11:41 am - +Rose M K Ho: Er ....
U still dun get wat I m saying.
Before we went so High Tech, we had No Computers !
Human intervention is necessary when everything else fails ... how long wl it take for things to get bk to normal if humans take back control ? -- tt is the back up.
Cyber ... still talking in tech terms , relying on computers.... not wat I'm thinking.
Hv u not seen enuff sci-fi movies like War of the Worlds (starring Tom Cruise), recently aired on Ch 5 .... ?
The way things r going, oyr world may become like tt one day.
Dun need aliens to invade our planet!
π±
19/07/2023, 11:41 am - +~L: We are not talking about Skynet and Terminators on the street (yet).
19/07/2023, 11:41 am - +Rose M K Ho: *yet*
... is the operative word
19/07/2023, 11:41 am - +~L: When we had no computers many more people died.
19/07/2023, 11:42 am - +Rose M K Ho: ** our world
19/07/2023, 11:42 am - +Rose M K Ho: Lol.
We dun nd computers for tt.
Covid did its job ... sorry for the pun.
19/07/2023, 11:42 am - +~L: lol yes. When that is being proposed we might need to be more careful. Right now, traffic simulations to decide new routes for roads, or population simulations to decide on new malls seems pretty harmless.
19/07/2023, 11:43 am - +Frankie Wee: It’s the same like ATM when you withdraw or deposit into machine cash is a computer is useful mean to the future AI support to work on human
19/07/2023, 11:44 am - +Frankie Wee: Either you use PayNow money sent within a mins they received
19/07/2023, 11:44 am - +Rose M K Ho: They always start out like this ... _harmless_ ....
Then, after we get comfortable & reliant on the system .... ??? π³
19/07/2023, 11:44 am - +~L: I'm sure people didn't want Photocopiers invented because it put typists out of work and could potential change text and create chaos in our documents..?
19/07/2023, 11:45 am - +Frankie Wee: Do I believe AI can safeguard against scam or fraud blocked them away.
19/07/2023, 11:46 am - +~L: It already does. This is how Spam filters work.
19/07/2023, 11:46 am - +Frankie Wee: Developing project will successfully
19/07/2023, 11:46 am - +~L: "AI" has been in use for years.... behind the scenes.
19/07/2023, 11:46 am - +Rose M K Ho: Agree .
But AI = Artificial Intelligence
... in case anyone doesn't know wat it stands for.
AI is much much much more powerful than all these other inventions.
It is potentially dangerous if not _controlled_ properly.
19/07/2023, 11:46 am - +~L: Now it's being made accessible to the general public
19/07/2023, 11:47 am - +Rose M K Ho: Ther wl be a Dark Web somewhere tt wl harness its power for unethical uses.
19/07/2023, 11:47 am - +Rose M K Ho: R we ready for it ?
19/07/2023, 11:48 am - +Frankie Wee: Cisco Systems will protect away dark web
19/07/2023, 11:48 am - +~L: Casio Calculator in 1978 used AI... It appeared to use intelligence to do maths, but it's just ones and zeros.
19/07/2023, 11:48 am - +Rose M K Ho: Those r primitive now, in comparison.
Lol.
19/07/2023, 11:50 am - +Frankie Wee: Without AI what will happen in Singapore? Other country developed AI fast maybe one day our CPF lose without support AI system.
19/07/2023, 11:51 am - +~L: We cannot stop the use of modern AI. It cannot be regulated. AI will keep going and will infiltrate every part of our lives. We are best positioned to take advantage of it, and ensure Singapore remains at the forefront of advanced technology for the good of the population.
19/07/2023, 11:53 am - +Rose M K Ho: Nah!
We r moving too fast !
Other countries in SE Asia ...? Dun think they r even close to this yet, as far as general public r concerned.
As for countries outside of us, tech is not readily adopted by _everyone_, esp if the country is very big and spread over many islands / hv many remote areas tt still hv ppl struggling with their daily living.
Tech is not for everyone.
19/07/2023, 11:54 am - +~L: It will be imposed on everyone. It will be impossible to avoid, unless people want to opt out and live in a hut growing vegetables.
19/07/2023, 11:54 am - +Rose M K Ho: π€£π€£ hahaha!
19/07/2023, 11:55 am - +~L: I don't think we are comparing ourselves to other countries in SE Asia.
19/07/2023, 11:55 am - +Rose M K Ho: So, if we can't beat them, we join them !!
Right ?
π€ͺπ€ͺ
19/07/2023, 11:55 am - +~L: Yes - and we can lead the way in the development of Ethical AI.
19/07/2023, 11:56 am - +~L: Protect jobs by upskilling our people. Protect economy by remaining competitive. Protect our future by educating kids in the use of AI and lateral thinking.
19/07/2023, 11:56 am - +Rose M K Ho: "Ethical" ....
Back to my point on putting controls & policing policies in place ....
π
19/07/2023, 11:57 am - +~L: The days of remembering facts and creating engineering diagrams are gone, or will be in 10 years.
19/07/2023, 11:57 am - +~L: Those you cannot do, effectively. I mean Ethical as in protecting people from the consequences of AI.
19/07/2023, 11:58 am - +Rose M K Ho: Same thing... π
U need rules (policies) in order to know wher the line has bn crossed.
19/07/2023, 12:00 pm - +~L: Consequences as in job losses, inadequate compensation for training the model, new industries etc
19/07/2023, 12:15 pm - +Jimmy Chew: It's good to embrace new technology and they need to double check the outputs become I heard chatgpt us 60 percent accurate and like the earlier terms in computing.. garbage in garbage out. They should use it as a supplement not as actual like PowerPoint. Haha it also allows them to earn good pay without working or thinking hahahaha
19/07/2023, 12:16 pm - +Jimmy Chew: This message was deleted
19/07/2023, 12:18 pm - +Jimmy Chew: Not sure whether it will weaken their analogical, logical, deep thinking skills to create unique insights and ideas to bring us to the next frontier because it's a gatherer of information and consolidators
19/07/2023, 12:18 pm - +Jimmy Chew: It's extremely good at it
19/07/2023, 12:19 pm - +Jimmy Chew: I used chatGPT after awhile I dropped thinking and trying to master the prompts that give the best reply hahaha..make my life easier but my brain lazier
19/07/2023, 12:25 pm - +Jimmy Chew: Haha it's like in the old days pre-printed replies to letters from customers hahahaha hope that not the case in the civil service
19/07/2023, 12:26 pm - +Jimmy Chew: They would use it like R2D2 to add value
19/07/2023, 12:26 pm - +Jimmy Chew: Or C3PO
19/07/2023, 12:28 pm - +~L: it's just a different way of thinking - requires structure, lateral approach, communication skills. Like having an assistant on hand :)
19/07/2023, 12:36 pm - +Jimmy Chew: Yeap. Only challenge is we will find the easy way. Like if you have a maid. You would leave the shopping and menu planning for meals and then you can go play hahaha
19/07/2023, 12:38 pm - +Jimmy Chew: <Media omitted>
19/07/2023, 12:38 pm - +Jimmy Chew: Like this guy
19/07/2023, 12:42 pm - +lesli: For your lunchtime entertainment ..
WATCH LIVE: WP media conference at 1pm on Leon Perera and Nicole Seah’s ‘inappropriate exchange’
July 19, 2023 12:15 PM
SINGAPORE - Workers’ Party (WP) secretary-general Pritam Singh will be speaking to members of the press at 1pm on Wednesday to address a video circulating online of Aljunied GRC MP Leon Perera and senior member Nicole Seah touching hands.
Read more: https://bit.ly/3pYk3EY
19/07/2023, 12:44 pm - +lesli: https://www.youtube.com/live/v1Omcc7OQA8?feature=share
19/07/2023, 1:17 pm - +Jimmy Chew: <Media omitted>
19/07/2023, 1:17 pm - +Jimmy Chew: Rec'd from a friend
19/07/2023, 1:19 pm - +Rama: Double standards as usual!
19/07/2023, 1:27 pm - +Rose M K Ho: Also my point abt using AI & leaving it to the machines...
Our brains become useless... π€ͺ
19/07/2023, 1:32 pm - +Rama: https://www.asiaone.com/singapore/leon-perera-resigns-mp-and-workers-party-nicole-seah-also-quits-party-pritam-singh
19/07/2023, 1:47 pm - +Frankie Wee: Both of them are unpolitical
19/07/2023, 1:48 pm - +Rama: Agree
19/07/2023, 1:59 pm - +REACH: π’ *Topic* π’
Some 4,000 civil servants are using a Government chatbot based on ChatGPT called Pair to assist them with writing, research, and coding, in a trial of artificial intelligence (AI) bots to boost productivity within the Government and better benefit the public.
A security arrangement between the Government and Azure OpenAI – an arm of Microsoft that offers coding and language AI models – also ensures that confidential information is kept private within ministries and unavailable to Microsoft.
π¬ *What are your views on the application of AI within Government services? What are some other ways of AI application that could help boost productivity of Government services and better benefit the public?*
Pair and an upgraded SupportGoWhere are part of the first tranche of AI use cases in the government. Pair will allow the secure and efficient use of large language models as a writing assistant within the government space, and also provides “a sparring partner for ideas”, he said.
The agency is also in the nascent stages of testing large language models to upgrade SupportGoWhere, a portal for locals to search and apply for government aid. Incorporating AI into SupportGoWhere will allow users to describe their problem in their own words and get a response on what assistance they can apply for and where to get it.
“If we get this approach right,” said Dr Janil, “citizens will be able to express their personal story without necessarily knowing how the schemes and support of the government are structured, which is something you still have to know today.”
Members of the public can take advantage of certain ePrimer modules on the Singapore Government Developer Portal to pick up these general concepts. (Link below - the last link)
ππ» *4,000 civil servants using Government Pair chatbot for writing, coding*
https://str.sg/iwvS
ππ» *How Singapore intends to harness AI for the public good*
https://str.sg/i3Ms
ππ» *Public officers can use ChatGPT and similar AI, but must take responsibility for their work: MCI*
https://str.sg/iJ8c
ππ» *Link to the Singapore Government Developer Portal*
https://www.developer.tech.gov.sg/
19/07/2023, 2:03 pm - +~L: Having used ChatGPT extensively, I concluded that you cannot trust it for factual information. It's a language model only. It can perform analysis of text. You cannot use it verbatim. It's a very useful tool, like an auto spell checker.... It cannot do the work for you, even though it appears it can.
19/07/2023, 2:04 pm - +~L: It can break down information quickly into simple terms and constraints. It can create graphs, write code, suggest options for you.
19/07/2023, 2:05 pm - +~L: It can write text explaining ideas and adjust the tone for different audiences to make it easier to understand.
19/07/2023, 2:06 pm - +~L: It's truly brilliant tech and very easy to use. However, it doesn't replace human experience. The people at risk are interns and junior data analysts for example.
19/07/2023, 2:07 pm - +~L: There are countless examples of where AI has been extremely productive and efficient doing the work of many people in a fraction of the time and at zero cost.
19/07/2023, 2:07 pm - +~L: We cannot afford not to adopt it.
19/07/2023, 2:09 pm - +Jimmy Chew: Maybe they can tap the top civil service thinking algorithms and program it. All the thinking process into a super thinkers
19/07/2023, 2:10 pm - +Rose M K Ho: Ya.... π€ͺ Then we dun need Super thinkers anymore ... ya?!
Cos super thinker dun need to use brain anymore, become lazy ... π΄
19/07/2023, 2:12 pm - +~L: https://chat.openai.com/share/1b16fd23-cfd7-4ef4-a253-4832c5f964b3
19/07/2023, 2:12 pm - +~L: Reach can start to use OpenAI ChatGPT to ask the same questions they are asking us each day :)
19/07/2023, 2:13 pm - +Rama: Using AI rather humans¡?
19/07/2023, 2:13 pm - +~L: The above example shows how generic the ChatBot is. However, it does provide structure that you can then drill down into and get more specific.
19/07/2023, 2:14 pm - +~L: Use AI to create specific structure and questions and suggestions - then ask the humans. AI does 80% of the work, which the humans then validate.
19/07/2023, 2:17 pm - +Rose M K Ho: Thanks!
It's gg to take a lot more than this list to convince me !
Btw, let's also see how 'sustainable' this is.
In the 1st place, they'll hv to deal w those who aren't even tech savvy ... using mobile phone is alredi a hassle for some - altho I doubt if this grp is the target audience.
"Eh" (A)? "ai / ei η±" (I) ?
Wat?
π€£π€£π€£π€£
19/07/2023, 2:17 pm - +Ah Heng: dun forget whatever AI is doing it still needs a man there to mgt it
19/07/2023, 2:18 pm - +Ah Heng: but many brainies from the west have came out also to condem the going to over reliance on AI where we humans allow it to be smarter than us leading to a very dark world if not well control
19/07/2023, 2:18 pm - +Ah Heng: some call it skynet for short
19/07/2023, 2:21 pm - +~L: Yes, of course. But it took 5 seconds to create that list. Doing it properly would take about an hour. Doing it without the use of AI would take at least a day.
19/07/2023, 2:23 pm - ☸️ Danny εΏ:
Prompt: Write a 600-word essay for a university scholarship application into a machine learning university course in a perspective of a researcher.
19/07/2023, 2:23 pm - ☸️ Danny εΏ:
As a researcher with a passion for machine learning, I am eager to apply for the machine learning scholarship at your esteemed university. I am confident that this scholarship will provide me with the opportunity to gain the technical skills and knowledge necessary to advance my research in the field.
My research interests lie in the application of machine learning to solve complex problems in various industries, such as healthcare and finance. While I have gained a solid foundation in machine learning through my previous academic and research experiences, I recognize that there is still much to learn. I am particularly interested in gaining deeper knowledge of advanced techniques such as deep learning and natural language processing, as well as the practical skills required to implement these techniques effectively.
I believe that the machine learning program at your university offers an exceptional opportunity to learn these skills. The program's curriculum is designed to provide students with a strong foundation in the fundamental concepts of machine learning while also providing hands-on experience with real-world datasets and problems. Additionally, the program offers courses on advanced techniques such as deep learning, natural language processing, and computer vision, which I believe will be invaluable in advancing my research.
In addition to the technical skills I hope to gain from this program, I am excited about the opportunity to collaborate with fellow researchers and learn from leading experts in the field. I believe that collaboration is essential in the field of machine learning, as it can lead to innovative solutions to complex problems. I am eager to work with other researchers on projects that apply machine learning to solve real-world problems.
Furthermore, I am excited about the possibility of participating in research projects conducted at your university. I believe that working on research projects will provide me with practical experience in implementing machine learning techniques and working with real-world datasets. Additionally, I am eager to contribute to ongoing research projects and learn from experienced researchers who have already made significant contributions to the field.
Lastly, I would like to mention that I am committed to using my technical skills and knowledge to make a positive impact on society. I believe that technology has the power to solve many of the world's most pressing problems, and I am passionate about contributing to this cause. After completing my studies, I plan to work on projects that use machine learning to address social problems such as healthcare disparities, financial inequality, and climate change.
In conclusion, I am confident that the machine learning program at your university will provide me with the technical skills and knowledge necessary to advance my research in the field. I am excited about the opportunity to collaborate with fellow researchers and learn from leading experts in the field. Additionally, I am committed to using my skills and knowledge to make a positive impact on society, and I believe that the skills I will gain from this program will enable me to do so effectively. Thank you for considering my application for this scholarship, and I look forward to the opportunity to contribute to the machine learning community at your university.
19/07/2023, 2:31 pm - +Frankie Wee: NASA will use AI to sent mission in mars planet than human
19/07/2023, 2:31 pm - ☸️ Danny εΏ:
1. To discuss the use of Generative AI, eg. ChatGPT, using a real life example will be the best to convince people who have not used it.
2. The above is an example of a prompt in ChatGPT and the response from ChatGPT.
3. The response look impressive.
4. But if submitted, the person will probably not successful in winning the scholarship.
5. It only provides a good report structure - but have no meat, specific ingredients that interviewers are looking for - how the applicant will benefit the organisation if scholarship is awarded to him.
6. So actually, there is no fear this generative AI will take over the job of the people.
It provides a good guide, a good structure to report writing, but it can't replace a human input.
7. To be honest, the final report is more than 80% different from this ChatGPT generated report when submitted.
8. And he was awarded the scholarship.
19/07/2023, 2:31 pm - +Frankie Wee: AI will building project in mars once completed mission we can live in mars
19/07/2023, 2:31 pm - +~L: Don't forget the prompt to write in the style of an enthusiastic student and not a GPT LLM.
19/07/2023, 2:32 pm - +Rose M K Ho: ππ»er ...
U mean this is Example generated by AI, right?
Yup. I'm fully aware of its capabilities.
19/07/2023, 2:32 pm - +Rose M K Ho: But I Dun Trust AI.... π€ͺ
19/07/2023, 2:33 pm - ☸️ Danny εΏ:
Yes it is generated by ChatGPT.
19/07/2023, 2:34 pm - +Frankie Wee: Wow you use ChatHPT?
19/07/2023, 2:34 pm - ☸️ Danny εΏ:
But it didn't provide insights how it benefit the organisation specifically.
See the report - it will give you a feel that the report is a "motherhood statement".
19/07/2023, 2:35 pm - +Rose M K Ho: I beg to differ.
'Lazy' people, Not so English-versed people wl _actually_ use this sample as it is without gg thru it!
That's _my_ point!
Not everyone wud look thru such verbose stuff w a fine tooth comb .... π
19/07/2023, 2:35 pm - +Frankie Wee: ChatGPT write within 5-10 seconds fast than human brain
19/07/2023, 2:35 pm - ☸️ Danny εΏ:
I too old to apply for a scholarship.
19/07/2023, 2:36 pm - +lesli: Perera's & Nicole's letters about their resignation & Pritam Singh's replies. From the horse's mouth.
Perera's letter https://bit.ly/3XWf2JA
Seah's letter https://bit.ly/3K6ny33
19/07/2023, 2:36 pm - +~L: Yes, exactly. It's a suggestion and a guide rather than to be used literally.
19/07/2023, 2:44 pm - +#.@: lacking in anecdotal evidence, didn't provide examples of ML project the candidate attempted on or libraries that was used, or interest in the university specific experiment, this will be brush aside if its a real application
19/07/2023, 2:45 pm - ☸️ Danny εΏ:
Yes.
Hence ChatGPT is not a real threat.
Just like computer, it is a tool not a replacement of human input. <This message was edited>
19/07/2023, 2:46 pm - +#.@: the real current threat is when your competitors are responding faster and cheaper by using AI
19/07/2023, 2:46 pm - +Rama: More understanding and legislation needed.
19/07/2023, 2:47 pm - ☸️ Danny εΏ:
Yes.
We need AI to hasten output by helping us.
Just like the invention of computer versus typewriter.
19/07/2023, 2:49 pm - +Frankie Wee: Time is changed
19/07/2023, 2:50 pm - +Frankie Wee: Like you use English dictionary book in old day searching
19/07/2023, 2:52 pm - ☸️ Danny εΏ:
Eg.
1. Computer versus typewriter.
2. Calculator versus abacus.
3. Machine versus handmade products.
19/07/2023, 2:53 pm - ☸️ Danny εΏ:
4. Aircon, fans versus hand fan.
19/07/2023, 2:53 pm - ☸️ Danny εΏ:
5. Car versus horse, ox cart.
19/07/2023, 2:54 pm - +Frankie Wee: Auto Machine wash versus hand wash
19/07/2023, 2:55 pm - ☸️ Danny εΏ:
Hahaha...
Yes.
AI is of similar vein.
19/07/2023, 2:57 pm - +Smiley face: 19 July, 2023
"Seeing is believing...?"
" Imagining the future of prepping for examinations; parents using chatbots to sieve the top probable maths questions for the PSLE or your tutor preparing a host of questions for your coming "A" levels exams or SEAB commanding the chatbots to derive, to repeat and to refine a sets of papers for the next national exams whether is a language, science, maths or theory of knowledge papers and no more tears, no more fears and less complaints; that's efficiency and how effective will all these faster setups becoming to be ACED by many people; after all, it is a game of ratio?"
- - anonymity
19/07/2023, 3:03 pm - +~L: I spoke with a legal firm and they said ChatGPT did a better job than their Juniors, so they will be hiring less Juniors.
19/07/2023, 3:03 pm - +~L: Does anyone here not use GPS via their Maps App on their Phone?
19/07/2023, 3:04 pm - ☸️ Danny εΏ:
US lawyer sorry after ChatGPT creates 'bogus' cases. Cannot fully trust ChatGPT while exploiting it. Must verify all the facts produced. https://www.channelnewsasia.com/world/us-lawyer-sorry-after-chatgpt-creates-bogus-cases-3549261
19/07/2023, 3:04 pm - +Rose M K Ho: Personally, I think everyone is _under_ estimating the potential of AI here.
U're not seeing wat I see.
Ther r always 2 sides to a coin.
Here, I see the Bad side of AI -- being misused for whatever (wl not mention since no one here understands wat I'm trying to say).
That wud be in the future, which I hope , will never happen.
But until tt day comes, it's good to see u guys here think so naively of the situation.
I'm looking way, way beyond all the "goodie" things tt everyone is "wowing" over.
It's no big deal, as it is , for now.
But its potential has yet to be reached & until someone 'sees' tt , or some super event blows up in our faces, I guess it's ok to just stick to the menial, trivial tasks tt everyone thinks AI can only do at the moment.
19/07/2023, 3:06 pm - +Caleb: fake URL?
19/07/2023, 3:06 pm - +Rose M K Ho: And yes, I see the potential Threat in AI, not the 'nice, wonderful' things everyone is wowing over.
19/07/2023, 3:06 pm - ☸️ Danny εΏ:
Hence though ChatGPT is a good tool, facts, checks, human inputs are imperative.
We cannot submit a ChatGPT generated output and submit it log stock and barrel.
The outcome could be exactly the opposite.
19/07/2023, 3:06 pm - +Caleb: Page Not Found
19/07/2023, 3:07 pm - ☸️ Danny εΏ:
Oh CNA remove the news already.
19/07/2023, 3:08 pm - ☸️ Danny εΏ:
"Lies, damn lies and ChatGPT"
https://www.businesstimes.com.sg/opinion-features/lies-damn-lies-and-chatgpt#:~:text=Lies%2C%20damn%20lies%20and%20ChatGPT
19/07/2023, 3:11 pm - ☸️ Danny εΏ:
This lawyer was sanction by the judge and law society - because ChatGPT generated fake case precedents .
The lawyer did not check the facts.
When the judge search his database for all the case precedents quoted, none was found.
ChatGPT fabricated all the case precedents.
If the case precedents are true, the lawyer could have win the case.
But now the lawyer got sanction and lose his case.
19/07/2023, 3:13 pm - +Rose M K Ho: Tks! Another good example ... this time of how AI can be misused ...
.. which brings me bk to the point of checks & policies ...
Had the judge not checked, .... wat wud hv happened.... ?
π€
19/07/2023, 3:16 pm - +#.@: For me the questions here are, did ChatGPT generate fake precedents, or did the lawyer wanted chatGPT to generate fake precedents. Did ChatGPT return fake results without the prompt by the lawyer?
19/07/2023, 3:17 pm - +#.@: but what would I know, i'm just naΓ―ve π₯²
19/07/2023, 3:18 pm - ☸️ Danny εΏ:
No specific on how the lawyer use the prompt.
The judge discovered that the lawyer use it honestly.
19/07/2023, 3:19 pm - ☸️ Danny εΏ:
But ChatGPT is trained on data provided by humans.
Garbage in garbage out.
ChatGPT could have pick up the fake case precedents provided in the internet, being trained on it and Garbage out.
19/07/2023, 3:20 pm - +Smiley face: Hi Rose, great analogy of AI...
The US-UK-EU are currently working towards a consensus of regulations for the use and future uses of AI, that's work in progress!
Second, you are correct! The good, the Bad and the Ugly of the coming of the AI world!
Third, only human as the creator of these powerful tools are responsible to not overly tune it till there may be no point of returning to the norm!
As Thomas Friedman once said, " the world is flat" and today it seems more flattering...
Thank you!
19/07/2023, 3:20 pm - +Rose M K Ho: <Media omitted>
19/07/2023, 3:20 pm - ☸️ Danny εΏ:
Also there is a possibility that AI do experience "hallucinations" - whereby algorithm go out of loop and produce gibberish output.
19/07/2023, 3:21 pm - +Rose M K Ho: There you go !
Machine gets the better of man ! ....
I rest my case!
π
19/07/2023, 3:23 pm - ☸️ Danny εΏ: <Media omitted>
19/07/2023, 3:24 pm - ☸️ Danny εΏ: <Media omitted>
19/07/2023, 3:25 pm - ☸️ Danny εΏ: https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)#:~:text=For%20example%2C%20a%20hallucinating%20chatbot,plucked%20from%20thin%20air%22).
19/07/2023, 3:28 pm - ☸️ Danny εΏ: <Media omitted>
19/07/2023, 3:30 pm - ☸️ Danny εΏ:
But highlighting the problems of ChatGPT - doesn't mean that I discourage using it.
We should use it.
But always check, verify, use critical analysis, and most important, human inputs and brainwork are very important.
Don't fully rely on ChatGPT to do the work for you.
Just use it as a tool for reference.
19/07/2023, 3:32 pm - +Rose M K Ho: Well said. Agreed.
19/07/2023, 3:47 pm - +~L: We agree. Exactly what I said in my earlier post too.
19/07/2023, 3:52 pm - ☸️ Danny εΏ:
1. Looking at the narrative of today's topic, the Government seems to adopt a "closed architecture" in exploiting an equivalent ChatGPT with Microsoft.
2. Look like data set being trained are not lifted from the open Internet - but within the civil service.
3. This will minimise the risk of fake data being inputted and trained on by Generative AI.
4. However, check, verification, human inputs and critical thinking will still be critical - as we cannot fully rely on Generative AI log-stock-and-barrel without fact checking and critical human thinking and input.
5. The mindset of using Generative AI - should be one like "using computer to help us" and not let "computer to replace us".
In similar vein, let "Generative AI to help us" and not let "Generative AI to replace us".
19/07/2023, 3:55 pm - +Smiley face: Code of Conduct?
19/07/2023, 3:56 pm - +Smiley face: Scale: The ratio of efficiency between the multiple points of services and the Servers.
19/07/2023, 3:56 pm - +Smiley face: Definitions and scopes of services?
19/07/2023, 3:57 pm - +Smiley face: Local, regional and global?
19/07/2023, 3:57 pm - +Smiley face: Securitize and Transactional?
19/07/2023, 3:58 pm - +Smiley face: How AI and chatbots will command the faculty of disinformation; if an entity pre-planned to use these digitised tools just to win?
19/07/2023, 3:58 pm - +Frankie Wee: Lawyer can use ChatGPT or ChatBox for battle win to win in case their clients
19/07/2023, 4:00 pm - +Smiley face: Created multiple precedents or new ways in the interpretations of the law, aka rewriting the law, once and repeatedly!
19/07/2023, 4:00 pm - +REACH: π’ *Topic* π’
Some 4,000 civil servants are using a Government chatbot based on ChatGPT called Pair to assist them with writing, research, and coding, in a trial of artificial intelligence (AI) bots to boost productivity within the Government and better benefit the public.
A security arrangement between the Government and Azure OpenAI – an arm of Microsoft that offers coding and language AI models – also ensures that confidential information is kept private within ministries and unavailable to Microsoft.
π¬ *What are your views on the application of AI within Government services? What are some other ways of AI application that could help boost productivity of Government services and better benefit the public?*
Pair and an upgraded SupportGoWhere are part of the first tranche of AI use cases in the government. Pair will allow the secure and efficient use of large language models as a writing assistant within the government space, and also provides “a sparring partner for ideas”, he said.
The agency is also in the nascent stages of testing large language models to upgrade SupportGoWhere, a portal for locals to search and apply for government aid. Incorporating AI into SupportGoWhere will allow users to describe their problem in their own words and get a response on what assistance they can apply for and where to get it.
“If we get this approach right,” said Dr Janil, “citizens will be able to express their personal story without necessarily knowing how the schemes and support of the government are structured, which is something you still have to know today.”
Members of the public can take advantage of certain ePrimer modules on the Singapore Government Developer Portal to pick up these general concepts. (Link below - the last link)
ππ» *4,000 civil servants using Government Pair chatbot for writing, coding*
https://str.sg/iwvS
ππ» *How Singapore intends to harness AI for the public good*
https://str.sg/i3Ms
ππ» *Public officers can use ChatGPT and similar AI, but must take responsibility for their work: MCI*
https://str.sg/iJ8c
ππ» *Link to the Singapore Government Developer Portal*
https://www.developer.tech.gov.sg/
19/07/2023, 4:04 pm - +Smiley face: Streamline and efficiency of the civil service to cater to thousands of users within minutes of delivering the specific service or enquiry with minimal Q-time wasted...
19/07/2023, 4:07 pm - +Rama: I heard la but don't know true anot that in 2025 GE, There will be a new party..Party members are Nicole Seah, Leon, TCJ and CLH, Palmer and David.
Party name have been decided ..Is call The Third Party. ππ€£π€£π€£
19/07/2023, 4:11 pm - +Ah Heng: wahhhhhh
19/07/2023, 4:32 pm - +Smiley face: https://youtu.be/j0rxzgoG7nU
19/07/2023, 4:33 pm - +Smiley face: https://youtu.be/Kmi7pDIk5N4
19/07/2023, 4:34 pm - +Rama: Only good contender against Tharman
19/07/2023, 4:36 pm - +Frankie Wee: He is former GIC
19/07/2023, 4:45 pm - ☸️ Danny εΏ:
1. To understand Generative AI, specifically, we have to understand what ChatGPT is made up of.
2. Generative AI, speficically use a combination of :-
a. ML - Machine Learning
b. NLP - Natural Language Processing
c. Deep Learning
d. LLM - Large Language Model.
e. Neural network
f. Reinforcement Learning
3. ML - Machine Learning
- is about creating a model that can recognise patterns in data and used them to make predictions or take actions.
4. NLP - Natural Language Processing
- enable machine to understand and process human language
5. Deep Learning
- It is based on deep learning architecture called the transformer.
6. LLM - Large Language Model.
- It mimic human intelligence.
- It uses statistical models to analyse vast amount of data, learning the patterns and connections between words and phrase.
7. Neural Network
- it uses neural network architecture comprising mathematical equations - linear and non-linear equations.
- A neural network comprise of multiple layers, where each layer performs a linear or non-linear transformation of the input from the previous layer.
8. Reinforcement Learning
- Reinforcement learning is a semi-supervised learning - learn based on "reward" (positive and negative reinforcement) - just like human.
- It is iterative in nature with the best solution gradually improving over time until an optimal strategy is discovered or the training process is ended.
- Because AI got "punished" with point deduction, it move away from bad solution and move into good solution as good solution accumulate points.
19/07/2023, 4:47 pm - ☸️ Danny εΏ: https://www.independent.co.uk/tech/harvard-chatbot-teacher-computer-science-b2363114.html
Harvard’s new computer science teacher is a chatbot
Students enrolled in the university’s flagship CS50 course will be presented with the AI teacher in September
19/07/2023, 4:48 pm - +Smiley face: https://youtu.be/sBWSszyI-GU
19/07/2023, 4:55 pm - +Smiley face: https://youtu.be/AaLQzXVL8oE
19/07/2023, 5:33 pm - +Smiley face: https://youtu.be/bSdUfnwQA8k
19/07/2023, 5:58 pm - +Rama: And PAP has changed its name to Problem After Problem party π€£
19/07/2023, 6:00 pm - +Frankie Wee: He is power political
19/07/2023, 6:00 pm - +REACH: π’ *Topic* π’
Some 4,000 civil servants are using a Government chatbot based on ChatGPT called Pair to assist them with writing, research, and coding, in a trial of artificial intelligence (AI) bots to boost productivity within the Government and better benefit the public.
A security arrangement between the Government and Azure OpenAI – an arm of Microsoft that offers coding and language AI models – also ensures that confidential information is kept private within ministries and unavailable to Microsoft.
π¬ *What are your views on the application of AI within Government services? What are some other ways of AI application that could help boost productivity of Government services and better benefit the public?*
Pair and an upgraded SupportGoWhere are part of the first tranche of AI use cases in the government. Pair will allow the secure and efficient use of large language models as a writing assistant within the government space, and also provides “a sparring partner for ideas”, he said.
The agency is also in the nascent stages of testing large language models to upgrade SupportGoWhere, a portal for locals to search and apply for government aid. Incorporating AI into SupportGoWhere will allow users to describe their problem in their own words and get a response on what assistance they can apply for and where to get it.
“If we get this approach right,” said Dr Janil, “citizens will be able to express their personal story without necessarily knowing how the schemes and support of the government are structured, which is something you still have to know today.”
Members of the public can take advantage of certain ePrimer modules on the Singapore Government Developer Portal to pick up these general concepts. (Link below - the last link)
ππ» *4,000 civil servants using Government Pair chatbot for writing, coding*
https://str.sg/iwvS
ππ» *How Singapore intends to harness AI for the public good*
https://str.sg/i3Ms
ππ» *Public officers can use ChatGPT and similar AI, but must take responsibility for their work: MCI*
https://str.sg/iJ8c
ππ» *Link to the Singapore Government Developer Portal*
https://www.developer.tech.gov.sg/
19/07/2023, 6:25 pm - Your security code with ~ M@ni changed. Tap to learn more.
19/07/2023, 6:45 pm - +REACH: *Dear Contributors,*
⏰ We will be closing the chat in *15 minutes* ⏰
Thank you very much for being part of our WhatsApp chat and participating actively.
Goodnight!
Megan π <This message was edited>
19/07/2023, 6:52 pm - +Rama: <Media omitted>
19/07/2023, 6:59 pm - +Smiley face: 19 July, 2023
"Seeing is believing...?"
" Imagining the future of prepping for examinations; parents using chatbots to sieve the top probable maths questions for the PSLE or your tutor preparing a host of questions for your coming "A" levels exams or SEAB commanding the chatbots to derive, to repeat and to refine a sets of papers for the next national exams whether is a language, science, maths or theory of knowledge papers and no more tears, no more fears and less complaints; that's efficiency and how effective will all these faster setups becoming to be ACED by many people; after all, it is a game of ratio?"
- - anonymity
Usefulness?
Streamline and efficiency of the civil service to cater to thousands of users within minutes of delivering the specific service or enquiry with minimal Queueing time via Internet, on phone or in person.
How AI and chatbots will command the faculty of disinformation; if an entity pre-planned to use these digitised tools just to win? How?
Seeing is believing?
Today what you see in the Internet can be not real ! How about public opinions, propagandas or even sentiments during a period of heated elections? How to define and detect what is true?
Scale?
The ratio of efficiency between the multiple points of services and the Servers.
It is an achievable and powerful tool even to scale a business model upwards into millions of options and outcomes, that's the power of AI vs human abilities - the absolute competitions for repetitive and refinement of tangibles and intangibles.
Definitions and scopes of services?
From the physical counters to the remote access and the paperworks to the digital platforms of information.
Securitize and Transactional? How to internalise and externalise risks of misused or leaked data?
Local, regional and global? Are these systems of apps come with several cloud banks (modular) in chains (obital) and are these apps accessible only by location from local or to regional or global. These are scaling and security issues to be addressed.
Let us expand this exams and chatbots model into many areas of possibility, impossibility and utilities; that's what Thomas Friedman once said, "the world is flat" and now is flattering - One World!
Afterwords:
Generative AI has broad applications that will impact a wide range of workers, occupations, and activities. Unlike most advances in automation of the past, it is a machine of the mind affecting cognitive work. (deep learning)
From WW2 until the early 1970s, labor productivity grew at over 3% a year, more than doubling over the period, ushering in an era of prosperity for most Americans. In the early 1970s productivity growth slowed dramatically, rebounding in the 1990s and only to slow again since the early 2000s. (growth & productivity)
The recent advances in generative AI have been driven by progress in software, hardware, data collection, and growing amounts of investment in cutting-edge models.
Cutting-edge AI systems has been doubling every six months over the past decade. The capabilities of generative AI systems have grown in multiples, allowing to perform many tasks that used to be reserved for the thinking workers, such as writing "well-crafted" sentences, creating computer code, summarising articles, brainstorming ideas, organising plans, translating multiple languages, writing complex emails, and more...(multiplier effects)
- - in progress - -
19/07/2023, 7:03 pm - +REACH: *Dear Contributors,*
We will be closing the chat for today.
Thank you very much for being part of our WhatsApp chat and participating actively.
Goodnight!
Megan π
19/07/2023, 7:03 pm - ~ REACH Singapore changed this group's settings to allow only admins to send messages to this group
=====
No comments:
Post a Comment