Deepseek AI

[TT]

Chinese open source AI which is as good/better than Chatgpt... with a
fraction of cost, training & price. Most downloaded app.

Nasdaq is going to absolutely crash today, maybe longer... lots of US AI
supremacy in the prices.

I tried Deepseek R1 during the weekend, and compared it to three US AI
bots. I posed few questions on my investment portfolio... I think Grok
had most relatable & best length of reply. But all but Deepseek failed
the math question...

My math/logic question about my portfolio (based on image of funds &
their percentages) was:

---

Question 1. "What percentage of the stocks are Chinese stocks in total?
And what percentage of the bond funds are in the European bond markets
in total?"

CORRECT ANSWER on Chinese stocks:
Msci China: 11%
EM Asia: 0.29 x 5%

=12,45%

---

-Google Gemini:

Conclusion: 11% + EM Asia

*SO-SO*

---

-Microsoft Copilot:
14% + part of 6%
...So basically same but got the submitted percentages incorrect!

*POOR*

---

-Grok:

14%

*POOR*

---

-Deepseek:

Based on guesstimate of EM Asia being 35% Chinese stocks (actually
currently 29%)

12.75%

*OK*

...None of the four AI seemed unable to access the Em Asia fund homepage
for correct percentage, and all failed in different ways the European
bonds question.


**************

Question 2.

"What is the average bond duration in my portfolio (excluding short
duration money-market funds)"?

Correct answer (my guesstimate calculated in head): around 9-10 years.

...None of the AI's could access actual current data, but some
guesstimates were given...


---

-Gemini:
Approx 5-8 years

*Not good*

---

Copilot:
5-7 years

I had to point out that it forgot long bonds, then it said 10-15 years

*BAD*

---

-Grok:
Approx 4,89 years

First try failed on weighted average math, answered 3 years something.
On second try today, failing big on 25+ EU bonds duration, claiming it 5
years!?! (it's 21 years)

*BAD*

---

Deepseek:

*Average duration of bond funds (excl. money market): ~8.8 years.

*This reflects significant interest rate sensitivity due to the
ultra-long-duration Eurozone 25+ bond fund (20 years).

*Implications
Interest Rate Risk: A 1% rise in rates could decrease the value of these
bonds by ~8.8%.

*Portfolio Impact: The long-duration bonds provide stability in
deflationary environments but are risky if rates rise sharply.

*Recommendation
Reduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20-year
duration) and replace it with intermediate-term bonds (e.g., 5–7 years).

*Verify Durations: Check fund factsheets for precise durations
(assumptions may vary slightly).

Would you like help adjusting allocations to lower duration risk? 🔍

*GOOD*

[TT]

Post by TT
Chinese open source AI which is as good/better than Chatgpt... with a
fraction of cost, training & price. Most downloaded app.
Nasdaq is going to absolutely crash today, maybe longer... lots of US AI
supremacy in the prices.
I tried Deepseek R1 during the weekend, and compared it to three US AI
bots. I posed few questions on my investment portfolio... I think Grok
had most relatable & best length of reply. But all but Deepseek failed
the math question...
My math/logic question about my portfolio (based on image of funds &
---
Question 1. "What percentage of the stocks are Chinese stocks in total?
And what percentage of the bond funds are in the European bond markets
in total?"
Msci China: 11%
EM Asia: 0.29 x 5%
=12,45%
---
Conclusion: 11% + EM Asia
*SO-SO*
---
14% + part of 6%
...So basically same but got the submitted percentages incorrect!
*POOR*
---
14%
*POOR*
---
Based on guesstimate of EM Asia being 35% Chinese stocks (actually
currently 29%)
12.75%
*OK*
...None of the four AI seemed unable to access the Em Asia fund homepage
for correct percentage, and all failed in different ways the European
bonds question.
**************
Question 2.
"What is the average bond duration in my portfolio (excluding short
duration money-market funds)"?
Correct answer (my guesstimate calculated in head): around 9-10 years.
...None of the AI's could access actual current data, but some
guesstimates were given...
---
Approx 5-8 years
*Not good*
---
5-7 years
I had to point out that it forgot long bonds, then it said 10-15 years
*BAD*
---
Approx 4,89 years
First try failed on weighted average math, answered 3 years something.
On second try today, failing big on 25+ EU bonds duration, claiming it 5
years!?! (it's 21 years)
*BAD*
---
*Average duration of bond funds (excl. money market): ~8.8 years.
*This reflects significant interest rate sensitivity due to the ultra-
long-duration Eurozone 25+ bond fund (20 years).
*Implications
Interest Rate Risk: A 1% rise in rates could decrease the value of these
bonds by ~8.8%.
*Portfolio Impact: The long-duration bonds provide stability in
deflationary environments but are risky if rates rise sharply.
*Recommendation
Reduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20-year
duration) and replace it with intermediate-term bonds (e.g., 5–7 years).
*Verify Durations: Check fund factsheets for precise durations
(assumptions may vary slightly).
Would you like help adjusting allocations to lower duration risk? 🔍
*GOOD*

P.S. at 13 Finland time...

-Nasdaq 100 futures are down 4.5%
-Hang Seng futures up 3%
-US 10 year down from 4.62 to 4.51

[PeteWasLucky]

Chinese open source AI which is as good/better than Chatgpt... with a fraction of cost, training & price. Most downloaded app.Nasdaq is going to absolutely crash today, maybe longer... lots of US AI supremacy in the prices.I tried Deepseek R1 during the weekend, and compared it to three US AI bots. I posed few questions on my investment portfolio... I think Grok had most relatable & best length of reply. But all but Deepseek failed the math question...My math/logic question about my portfolio (based on image of funds & their percentages) was:---Question 1. "What percentage of the stocks are Chinese stocks in total? And what percentage of the bond funds are in the European bond markets in total?"CORRECT ANSWER on Chinese stocks:Msci China: 11%EM Asia: 0.29 x 5%=12,45%----Google Gemini:Conclusion: 11% + EM Asia*SO-SO*----Microsoft Copilot:14% + part of 6%...So basically same but got the submitted percentages incorrect!*POOR*----Grok:14%*POOR*----Deepseek:Based on guesstimate of EM Asia being 35% Chinese stocks (actually currently 29%)12.75%*OK*...None of the four AI seemed unable to access the Em Asia fund homepage for correct percentage, and all failed in different ways the European bonds question.**************Question 2."What is the average bond duration in my portfolio (excluding short duration money-market funds)"?Correct answer (my guesstimate calculated in head): around 9-10 years....None of the AI's could access actual current data, but some guesstimates were given...----Gemini:Approx 5-8 years*Not good*---Copilot:5-7 yearsI had to point out that it forgot long bonds, then it said 10-15 years*BAD*----Grok:Approx 4,89 yearsFirst try failed on weighted average math, answered 3 years something.On second try today, failing big on 25+ EU bonds duration, claiming it 5 years!?! (it's 21 years)*BAD*---Deepseek:*Average duration of bond funds (excl. money market): ~8.8 years.*This reflects significant interest rate sensitivity due to the ultra-long-duration Eurozone 25+ bond fund (20 years).*ImplicationsInterest Rate Risk: A 1% rise in rates could decrease the value of these bonds by ~8.8%.*Portfolio Impact: The long-duration bonds provide stability in deflationary environments but are risky if rates rise sharply.*RecommendationReduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20-year duration) and replace it with intermediate-term bonds (e.g., 5–7 years).*Verify Durations: Check fund factsheets for precise durations (assumptions may vary slightly).Would you like help adjusting allocations to lower duration risk? *GOOD*

Thanks for sharing, I need to take a deep look on this.

[Sawfish]

Post by TT
Chinese open source AI which is as good/better than Chatgpt... with a
fraction of cost, training & price. Most downloaded app.
Nasdaq is going to absolutely crash today, maybe longer... lots of US AI
supremacy in the prices.
I tried Deepseek R1 during the weekend, and compared it to three US AI
bots. I posed few questions on my investment portfolio... I think Grok
had most relatable & best length of reply. But all but Deepseek failed
the math question...
My math/logic question about my portfolio (based on image of funds &
---
Question 1. "What percentage of the stocks are Chinese stocks in total?
And what percentage of the bond funds are in the European bond markets
in total?"
Msci China: 11%
EM Asia: 0.29 x 5%
=12,45%
---
Conclusion: 11% + EM Asia
*SO-SO*
---
14% + part of 6%
...So basically same but got the submitted percentages incorrect!
*POOR*
---
14%
*POOR*
---
Based on guesstimate of EM Asia being 35% Chinese stocks (actually
currently 29%)
12.75%
*OK*
...None of the four AI seemed unable to access the Em Asia fund homepage
for correct percentage, and all failed in different ways the European
bonds question.
**************
Question 2.
"What is the average bond duration in my portfolio (excluding short
duration money-market funds)"?
Correct answer (my guesstimate calculated in head): around 9-10 years.
...None of the AI's could access actual current data, but some
guesstimates were given...
---
Approx 5-8 years
*Not good*
---
5-7 years
I had to point out that it forgot long bonds, then it said 10-15 years
*BAD*
---
Approx 4,89 years
First try failed on weighted average math, answered 3 years something.
On second try today, failing big on 25+ EU bonds duration, claiming it 5
years!?! (it's 21 years)
*BAD*
---
*Average duration of bond funds (excl. money market): ~8.8 years.
*This reflects significant interest rate sensitivity due to the
ultra-long-duration Eurozone 25+ bond fund (20 years).
*Implications
Interest Rate Risk: A 1% rise in rates could decrease the value of these
bonds by ~8.8%.
*Portfolio Impact: The long-duration bonds provide stability in
deflationary environments but are risky if rates rise sharply.
*Recommendation
Reduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20-year
duration) and replace it with intermediate-term bonds (e.g., 5–7 years).
*Verify Durations: Check fund factsheets for precise durations
(assumptions may vary slightly).
Would you like help adjusting allocations to lower duration risk? 🔍
*GOOD*

I read this opinion essay in Substack yesterday. FWIW, I think that the
directions that AI will take industrialized societies as described in
the essay has at 75% chance of happening. Maybe more.

https://www.aporiamagazine.com/p/yes-youre-going-to-be-replaced
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"To the average American or Englishman the very name of anarchy causes a
shudder, because it invariably conjures up a picture of a land
terrorized by low-browed assassins with matted beards, carrying bombs in
one hand and mugs of beer in the other. But as a matter of fact, there
is no reason whatever to believe that, if all laws were abolished
tomorrow, such swine would survive the day."

--H. L. Mencken
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[TT]

Post by Sawfish

Post by TT
Chinese open source AI which is as good/better than Chatgpt... with a
fraction of cost, training & price. Most downloaded app.
Nasdaq is going to absolutely crash today, maybe longer... lots of US
AI supremacy in the prices.
I tried Deepseek R1 during the weekend, and compared it to three US AI
bots. I posed few questions on my investment portfolio... I think Grok
had most relatable & best length of reply. But all but Deepseek failed
the math question...
My math/logic question about my portfolio (based on image of funds &
---
Question 1. "What percentage of the stocks are Chinese stocks in
total? And what percentage of the bond funds are in the European bond
markets in total?"
Msci China: 11%
EM Asia: 0.29 x 5%
=12,45%
---
Conclusion: 11% + EM Asia
*SO-SO*
---
14% + part of 6%
...So basically same but got the submitted percentages incorrect!
*POOR*
---
14%
*POOR*
---
Based on guesstimate of EM Asia being 35% Chinese stocks (actually
currently 29%)
12.75%
*OK*
...None of the four AI seemed unable to access the Em Asia fund
homepage for correct percentage, and all failed in different ways the
European bonds question.
**************
Question 2.
"What is the average bond duration in my portfolio (excluding short
duration money-market funds)"?
Correct answer (my guesstimate calculated in head): around 9-10 years.
...None of the AI's could access actual current data, but some
guesstimates were given...
---
Approx 5-8 years
*Not good*
---
5-7 years
I had to point out that it forgot long bonds, then it said 10-15 years
*BAD*
---
Approx 4,89 years
First try failed on weighted average math, answered 3 years something.
On second try today, failing big on 25+ EU bonds duration, claiming it
5 years!?! (it's 21 years)
*BAD*
---
*Average duration of bond funds (excl. money market): ~8.8 years.
*This reflects significant interest rate sensitivity due to the ultra-
long-duration Eurozone 25+ bond fund (20 years).
*Implications
Interest Rate Risk: A 1% rise in rates could decrease the value of
these bonds by ~8.8%.
*Portfolio Impact: The long-duration bonds provide stability in
deflationary environments but are risky if rates rise sharply.
*Recommendation
Reduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20-
year duration) and replace it with intermediate-term bonds (e.g., 5–7
years).
*Verify Durations: Check fund factsheets for precise durations
(assumptions may vary slightly).
Would you like help adjusting allocations to lower duration risk? 🔍
*GOOD*

I read this opinion essay in Substack yesterday. FWIW, I think that the
directions that AI will take industrialized societies as described in
the essay has at 75% chance of happening. Maybe more.
https://www.aporiamagazine.com/p/yes-youre-going-to-be-replaced

Very interesting article. Lots of it is going to happen I think. If AI
in the future (already) does most computer tasks better than human...

I have to say that I've felt a bit smug vs AI, especially when I've had
the opportunity to correct its math, which has been quite a few times.
Let's enjoy it while it lasts...

Anyway, Deepseek etc may bring AI on shoestring budget to everyone's
reach. Meaning that it's not just few operators with endless billions to
burn...

[Sawfish]

Post by TT

Post by Sawfish

Post by TT
Chinese open source AI which is as good/better than Chatgpt... with a
fraction of cost, training & price. Most downloaded app.
Nasdaq is going to absolutely crash today, maybe longer... lots of US
AI supremacy in the prices.
I tried Deepseek R1 during the weekend, and compared it to three US
AI bots. I posed few questions on my investment portfolio... I think
Grok had most relatable & best length of reply. But all but Deepseek
failed the math question...
My math/logic question about my portfolio (based on image of funds &
---
Question 1. "What percentage of the stocks are Chinese stocks in
total? And what percentage of the bond funds are in the European bond
markets in total?"
Msci China: 11%
EM Asia: 0.29 x 5%
=12,45%
---
Conclusion: 11% + EM Asia
*SO-SO*
---
14% + part of 6%
...So basically same but got the submitted percentages incorrect!
*POOR*
---
14%
*POOR*
---
Based on guesstimate of EM Asia being 35% Chinese stocks (actually
currently 29%)
12.75%
*OK*
...None of the four AI seemed unable to access the Em Asia fund
homepage for correct percentage, and all failed in different ways the
European bonds question.
**************
Question 2.
"What is the average bond duration in my portfolio (excluding short
duration money-market funds)"?
Correct answer (my guesstimate calculated in head): around 9-10 years.
...None of the AI's could access actual current data, but some
guesstimates were given...
---
Approx 5-8 years
*Not good*
---
5-7 years
I had to point out that it forgot long bonds, then it said 10-15 years
*BAD*
---
Approx 4,89 years
First try failed on weighted average math, answered 3 years something.
On second try today, failing big on 25+ EU bonds duration, claiming
it 5 years!?! (it's 21 years)
*BAD*
---
*Average duration of bond funds (excl. money market): ~8.8 years.
*This reflects significant interest rate sensitivity due to the
ultra- long-duration Eurozone 25+ bond fund (20 years).
*Implications
Interest Rate Risk: A 1% rise in rates could decrease the value of
these bonds by ~8.8%.
*Portfolio Impact: The long-duration bonds provide stability in
deflationary environments but are risky if rates rise sharply.
*Recommendation
Reduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20-
year duration) and replace it with intermediate-term bonds (e.g., 5–7
years).
*Verify Durations: Check fund factsheets for precise durations
(assumptions may vary slightly).
Would you like help adjusting allocations to lower duration risk? 🔍
*GOOD*

I read this opinion essay in Substack yesterday. FWIW, I think that
the directions that AI will take industrialized societies as described
in the essay has at 75% chance of happening. Maybe more.
https://www.aporiamagazine.com/p/yes-youre-going-to-be-replaced

Very interesting article. Lots of it is going to happen I think. If AI
in the future (already) does most computer tasks better than human...
I have to say that I've felt a bit smug vs AI, especially when I've had
the opportunity to correct its math, which has been quite a few times.
Let's enjoy it while it lasts...
Anyway, Deepseek etc may bring AI on shoestring budget to everyone's
reach. Meaning that it's not just few operators with endless billions to
burn...

Agreed.

There's an odd way to look at widespread, cheap access to AI, and it is
similar to doing away with a priesthood so that each person can access
god, himself directly. Attorney, doctors, financial advisors are hold
the priestly role in today's society.

Like the Reformation times.

Things like prostitution, direct crime should be the last to be
affected, if ever, and living in a 3rd world country--a shithole
country--people there will see few changes.

[PeteWasLucky]

Sawfish kirjoitti 27.1.2025 klo 17.27:> On 1/27/25 2:44 AM, TT wrote:>> Chinese open source AI which is as good/better than Chatgpt... with a >> fraction of cost, training & price. Most downloaded app.>>>> Nasdaq is going to absolutely crash today, maybe longer... lots of US >> AI supremacy in the prices.>>>> I tried Deepseek R1 during the weekend, and compared it to three US AI >> bots. I posed few questions on my investment portfolio... I think Grok >> had most relatable & best length of reply. But all but Deepseek failed >> the math question...>>>> My math/logic question about my portfolio (based on image of funds & >> their percentages) was:>>>> --->>>> Question 1. "What percentage of the stocks are Chinese stocks in >> total? And what percentage of the bond funds are in the European bond >> markets in total?">>>> CORRECT ANSWER on Chinese stocks:>> Msci China: 11%>> EM Asia: 0.29 x 5%>>>> =12,45%>>>> --->>>> -Google Gemini:>>>> Conclusion: 11% + EM Asia>>>> *SO-SO*>>>> --->>>> -Microsoft Copilot:>> 14% + part of 6%>> ...So basically same but got the submitted percentages incorrect!>>>> *POOR*>>>> --->>>> -Grok:>>>> 14%>>>> *POOR*>>>> --->>>> -Deepseek:>>>> Based on guesstimate of EM Asia being 35% Chinese stocks (actually >> currently 29%)>>>> 12.75%>>>> *OK*>>>> ...None of the four AI seemed unable to access the Em Asia fund >> homepage for correct percentage, and all failed in different ways the >> European bonds question.>>>>>> **************>>>> Question 2.>>>> "What is the average bond duration in my portfolio (excluding short >> duration money-market funds)"?>>>> Correct answer (my guesstimate calculated in head): around 9-10 years.>>>> ...None of the AI's could access actual current data, but some >> guesstimates were given...>>>>>> --->>>> -Gemini:>> Approx 5-8 years>>>> *Not good*>>>> --->>>> Copilot:>> 5-7 years>>>> I had to point out that it forgot long bonds, then it said 10-15 years>>>> *BAD*>>>> --->>>> -Grok:>> Approx 4,89 years>>>> First try failed on weighted average math, answered 3 years something.>> On second try today, failing big on 25+ EU bonds duration, claiming it >> 5 years!?! (it's 21 years)>>>> *BAD*>>>> --->>>> Deepseek:>>>> *Average duration of bond funds (excl. money market): ~8.8 years.>>>> *This reflects significant interest rate sensitivity due to the ultra- >> long-duration Eurozone 25+ bond fund (20 years).>>>> *Implications>> Interest Rate Risk: A 1% rise in rates could decrease the value of >> these bonds by ~8.8%.>>>> *Portfolio Impact: The long-duration bonds provide stability in >> deflationary environments but are risky if rates rise sharply.>>>> *Recommendation>> Reduce Long-Duration Exposure: Trim the Eurozone 25+ allocation (20- >> year duration) and replace it with intermediate-term bonds (e.g., 5–7 >> years).>>>> *Verify Durations: Check fund factsheets for precise durations >> (assumptions may vary slightly).>>>> Would you like help adjusting allocations to lower duration risk? >>>> *GOOD*> > I read this opinion essay in Substack yesterday. FWIW, I think that the > directions that AI will take industrialized societies as described in > the essay has at 75% chance of happening. Maybe more.> > https://www.aporiamagazine.com/p/yes-youre-going-to-be-replacedVery interesting article. Lots of it is going to happen I think. If AI in the future (already) does most computer tasks better than human...I have to say that I've felt a bit smug vs AI, especially when I've had the opportunity to correct its math, which has been quite a few times. Let's enjoy it while it lasts...Anyway, Deepseek etc may bring AI on shoestring budget to everyone's reach. Meaning that it's not just few operators with endless billions to burn...

In order to have AI fully overtake most of the human jobs for a long stable period of time, it must exist in an economy that is stable and thrives by its existence.
If the human population can't find jobs that pay enough in such a way they can drive consumer spending, then all the big companies that benefit from this AI will suffer and lose all their sales and profits.
Without that perfect balance, any of these models can't be sustained for long time.
But it definitely can happen for a short period of time, in which the people economy will be suffering and something big will have to happen to break the cycle including major wars.