StarZero All shows Try the podcast clipper →

Jan 15, 2015 · 26m · mad

Michael Karasick, IBM Watson // Data Driven #33 // Jan 2015 (Hosted by FirstMark Capital)

Michael Karasick · 17m spoken Matt Turck · 4m spoken Matt Kroll · 45s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

At a Data Driven NYC meetup hosted by FirstMark Capital, IBM VP of Watson R&D Michael Karasick outlines Watson's evolution from a Jeopardy!-winning quiz system into a cloud-based cognitive platform, detailing IBM's strategic investments, developer ecosystem, and healthcare applications.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 19.6% of the talking time here. How this is scored →

Matt as informed peer 2.0 Guest teaching 4.4 Guest disagreement 1.0 Matt pushing back 0.6
05100:0010:0020:002:06–4:33 · Matt as informed peer 2/10 The Watson Ecosystem and Venture Fund Matt asks about the stage and entry points for IBM's Watson ecosystem fund. Karasick clarifies that the fund focuses on Series B and C rounds rather than early-stage investments.4:33–9:31 · Matt as informed peer 1/10 Evolution of Watson from Jeopardy to Chef Watson Matt prompts Karasick on Watson's evolution from its Jeopardy origins. Karasick delivers an extended narrative covering the statistical precision graph required to beat human Jeopardy champions and the expansion into Chef Watson.9:31–16:16 · Matt as informed peer 4/10 Watson Architecture and Healthcare Applications Matt poses a theoretical question on horizontal versus vertical AI architectures, assuming Watson fits the horizontal model. Karasick gently reframes the premise, explaining that Watson operates as a network of specialized, domain-trained reasoners.16:16–21:50 · Matt as informed peer 3/10 Product Roadmap and Silicon Alley Headquarters Matt brings up IBM's major financial commitment to Watson and its new headquarters in Silicon Alley. Karasick elaborates on the Bluemix cloud services and hardware miniaturization.21:50–26:19 · Matt as informed peer 0/10 Audience Q&A and Event Conclusion Audience members ask technical questions regarding Watson replacing traditional big data stacks like Hadoop and why NLP was chosen. Karasick details IBM's historical focus on unstructured data and UIMA standards.2:06–4:33 · Guest teaching 3/10 The Watson Ecosystem and Venture Fund Matt asks about the stage and entry points for IBM's Watson ecosystem fund. Karasick clarifies that the fund focuses on Series B and C rounds rather than early-stage investments.4:33–9:31 · Guest teaching 5/10 Evolution of Watson from Jeopardy to Chef Watson Matt prompts Karasick on Watson's evolution from its Jeopardy origins. Karasick delivers an extended narrative covering the statistical precision graph required to beat human Jeopardy champions and the expansion into Chef Watson.9:31–16:16 · Guest teaching 6/10 Watson Architecture and Healthcare Applications Matt poses a theoretical question on horizontal versus vertical AI architectures, assuming Watson fits the horizontal model. Karasick gently reframes the premise, explaining that Watson operates as a network of specialized, domain-trained reasoners.16:16–21:50 · Guest teaching 4/10 Product Roadmap and Silicon Alley Headquarters Matt brings up IBM's major financial commitment to Watson and its new headquarters in Silicon Alley. Karasick elaborates on the Bluemix cloud services and hardware miniaturization.21:50–26:19 · Guest teaching 4/10 Audience Q&A and Event Conclusion Audience members ask technical questions regarding Watson replacing traditional big data stacks like Hadoop and why NLP was chosen. Karasick details IBM's historical focus on unstructured data and UIMA standards.2:06–4:33 · Guest disagreement 1/10 The Watson Ecosystem and Venture Fund Matt asks about the stage and entry points for IBM's Watson ecosystem fund. Karasick clarifies that the fund focuses on Series B and C rounds rather than early-stage investments.4:33–9:31 · Guest disagreement 1/10 Evolution of Watson from Jeopardy to Chef Watson Matt prompts Karasick on Watson's evolution from its Jeopardy origins. Karasick delivers an extended narrative covering the statistical precision graph required to beat human Jeopardy champions and the expansion into Chef Watson.9:31–16:16 · Guest disagreement 2/10 Watson Architecture and Healthcare Applications Matt poses a theoretical question on horizontal versus vertical AI architectures, assuming Watson fits the horizontal model. Karasick gently reframes the premise, explaining that Watson operates as a network of specialized, domain-trained reasoners.16:16–21:50 · Guest disagreement 0/10 Product Roadmap and Silicon Alley Headquarters Matt brings up IBM's major financial commitment to Watson and its new headquarters in Silicon Alley. Karasick elaborates on the Bluemix cloud services and hardware miniaturization.21:50–26:19 · Guest disagreement 1/10 Audience Q&A and Event Conclusion Audience members ask technical questions regarding Watson replacing traditional big data stacks like Hadoop and why NLP was chosen. Karasick details IBM's historical focus on unstructured data and UIMA standards.2:06–4:33 · Matt pushing back 1/10 The Watson Ecosystem and Venture Fund Matt asks about the stage and entry points for IBM's Watson ecosystem fund. Karasick clarifies that the fund focuses on Series B and C rounds rather than early-stage investments.4:33–9:31 · Matt pushing back 0/10 Evolution of Watson from Jeopardy to Chef Watson Matt prompts Karasick on Watson's evolution from its Jeopardy origins. Karasick delivers an extended narrative covering the statistical precision graph required to beat human Jeopardy champions and the expansion into Chef Watson.9:31–16:16 · Matt pushing back 2/10 Watson Architecture and Healthcare Applications Matt poses a theoretical question on horizontal versus vertical AI architectures, assuming Watson fits the horizontal model. Karasick gently reframes the premise, explaining that Watson operates as a network of specialized, domain-trained reasoners.16:16–21:50 · Matt pushing back 0/10 Product Roadmap and Silicon Alley Headquarters Matt brings up IBM's major financial commitment to Watson and its new headquarters in Silicon Alley. Karasick elaborates on the Bluemix cloud services and hardware miniaturization.21:50–26:19 · Matt pushing back 0/10 Audience Q&A and Event Conclusion Audience members ask technical questions regarding Watson replacing traditional big data stacks like Hadoop and why NLP was chosen. Karasick details IBM's historical focus on unstructured data and UIMA standards.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 32.2% · guest 67.8%0:00 · Matt 32.2% · guest 67.8%3:00 · Matt 33.3% · guest 66.7%3:00 · Matt 33.3% · guest 66.7%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 28% · guest 72%9:00 · Matt 28% · guest 72%12:00 · Matt 23.9% · guest 76.1%12:00 · Matt 23.9% · guest 76.1%15:00 · Matt 2.7% · guest 97.3%15:00 · Matt 2.7% · guest 97.3%18:00 · Matt 21.5% · guest 78.5%18:00 · Matt 21.5% · guest 78.5%21:00 · Matt 15.1% · guest 84.9%21:00 · Matt 15.1% · guest 84.9%24:00 · Matt 20% · guest 80%24:00 · Matt 20% · guest 80%
Sharpest disagreement ▶ 11:50 Friendly premise rejection on AI architecture

Karasick explicitly rejects Matt's assumption that Watson is a single horizontal global brain, explaining it is built on cooperating domain-specific reasoners.

 Hardest push from Matt ▶ 11:07 Framing the horizontal vs vertical AI debate

Matt frames a structured industry debate comparing general horizontal models against narrow vertical bots, forcing Karasick to take a stand on IBM's design philosophy.

 Biggest teaching moment ▶ 11:50 Reframing Watson's core system architecture

Karasick educates the host on how Watson combines roughly 100 specialized reasoners and scoring algorithms rather than operating as a monolithic engine.

 Matt holds his own ▶ 11:07 Articulating AI strategy taxonomies

Matt demonstrates high-level industry context by outlining the tension between single global AI engines and task-specific micro-bots.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
The Watson Ecosystem and Venture Fund 2311 Matt asks about the stage and entry points for IBM's Watson ecosystem fund. Karasick clarifies that the fund focuses on Series B and C rounds rather than early-stage investments.
Evolution of Watson from Jeopardy to Chef Watson 1510 Matt prompts Karasick on Watson's evolution from its Jeopardy origins. Karasick delivers an extended narrative covering the statistical precision graph required to beat human Jeopardy champions and the expansion into Chef Watson.
Watson Architecture and Healthcare Applications 4622 Matt poses a theoretical question on horizontal versus vertical AI architectures, assuming Watson fits the horizontal model. Karasick gently reframes the premise, explaining that Watson operates as a network of specialized, domain-trained reasoners.
Product Roadmap and Silicon Alley Headquarters 3400 Matt brings up IBM's major financial commitment to Watson and its new headquarters in Silicon Alley. Karasick elaborates on the Bluemix cloud services and hardware miniaturization.
Audience Q&A and Event Conclusion 0410 Audience members ask technical questions regarding Watson replacing traditional big data stacks like Hadoop and why NLP was chosen. Karasick details IBM's historical focus on unstructured data and UIMA standards.

Statements from this episode (9)

Disclosure
IBM created a $100 million Watson investment fund
“And we created a hundred million dollar investment fund.”
Michael Karasick Jan 15, 2015 ▶ 2:35
Disclosure
IBM Watson has made three minority investments in startups
“And we've done, I guess, three minority investments in startups who have decided to work with us.”
Michael Karasick Jan 15, 2015 ▶ 2:40
Disclosure
IBM Watson venture fund targets Series B and C startups
“B and C, yeah. We want people to sort of, kind of know where, where their feet are.”
Michael Karasick Jan 15, 2015 ▶ 3:57
Assertion Partly supported
IBM Watson released a cookbook on Amazon with 50 AI-created recipes
“There's a cookbook out also, and it's available on Amazon. Done, done with the Institute of Culinary Education. There's kind of 50 Watson-created recipes.”
Michael Karasick Jan 15, 2015 ▶ 9:10
Assertion Not checkable as stated
Oncologists, not financial firms, first contacted IBM Watson after Jeopardy
“The first people to call us after the Watson game, we expected the standard, I should say, really, really interesting IBM Customers, financial houses, didn't call us. In fact, oncologists, cancer doctors did.”
Michael Karasick Jan 15, 2015 ▶ 10:36
Assertion Not checkable as stated
IBM Watson was engineered as modular specialized bots, not one model
“The right way to think of Watson is, And the engineering is, is a lot of cooperating little bots. So think of all of the Watson engines as having an NLP component which is trained according to the domain. So there isn't one Watson, there's a bunch of them.”
Michael Karasick Jan 15, 2015 ▶ 11:50
Insight
Deploying AI systems requires explicit evidence behind recommendations
“One of the things we learned about systems like Watson, if you're going to deploy it is people want to know why. You know, give it an answer, and they go, they want evidence.”
Michael Karasick Jan 15, 2015 ▶ 15:49
Assertion Not checkable as stated
IBM Watson hardware efficiency improved 1,700x since the Jeopardy setup
“Watson began life as a, 2300 power seven servers running a multi-processing version of Watson, all in memory, running on Linux. The original cluster is still in Yorktown, people still want to see it. We've improved it by a factor of about 1700, so now it's a l…”
Michael Karasick Jan 15, 2015 ▶ 19:33
Assertion Supported
IBM Watson used Apache UIMA and did not replace Hadoop
“I wouldn't assert that it replaces Hadoop. In fact, it's based on UEMA. It's an Apache project.”
Michael Karasick Jan 15, 2015 ▶ 23:14
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more